Authors: Victor Nascimento Ribeiro, Nina S. T. Hirata
Abstract: Video-based Automatic License Plate Recognition (ALPR) involves extracting vehicle license plate text information from video captures. Traditional systems typically rely heavily on high-end computing resources and utilize multiple frames to recognize license plates, leading to increased computational overhead. In this paper, we propose two methods capable of efficiently extracting exactly one frame per vehicle and recognizing its license plate characters from this single image, thus significantly reducing computational demands. The first method uses Visual Rhythm (VR) to generate time-spatial images from videos, while the second employs Accumulative Line Analysis (ALA), a novel algorithm based on single-line video processing for real-time operation. Both methods leverage YOLO for license plate detection within the frame and a Convolutional Neural Network (CNN) for Optical Character Recognition (OCR) to extract textual information. Experiments on real videos demonstrate that the proposed methods achieve results comparable to traditional frame-by-frame approaches, with processing speeds three times faster.
Authors: Ulindu De Silva, Leon Fernando, Kalinga Bandara, Rashmika Nawaratne
Abstract: The proliferation of video content production has led to vast amounts of data, posing substantial challenges in terms of analysis efficiency and resource utilization. Addressing this issue calls for the development of robust video analysis tools. This paper proposes a novel approach leveraging Generative Artificial Intelligence (GenAI) to facilitate streamlined video analysis. Our tool aims to deliver tailored textual summaries of user-defined queries, offering a focused insight amidst extensive video datasets. Unlike conventional frameworks that offer generic summaries or limited action recognition, our method harnesses the power of GenAI to distil relevant information, enhancing analysis precision and efficiency. Employing YOLO-V8 for object detection and Gemini for comprehensive video and text analysis, our solution achieves heightened contextual accuracy. By combining YOLO with Gemini, our approach furnishes textual summaries extracted from extensive CCTV footage, enabling users to swiftly navigate and verify pertinent events without the need for exhaustive manual review. The quantitative evaluation revealed a similarity of 72.8%, while the qualitative assessment rated an accuracy of 85%, demonstrating the capability of the proposed method.
Authors: Felix Krause, Timy Phan, Vincent Tao Hu, Bj\"orn Ommer
Abstract: Diffusion models have emerged as the mainstream approach for visual generation. However, these models usually suffer from sample inefficiency and high training costs. This issue is particularly pronounced in the standard diffusion transformer architecture due to its quadratic complexity relative to input length. Recent works have addressed this by reducing the number of tokens processed in the model, often through masking. In contrast, this work aims to improve the training efficiency of the diffusion backbone by using predefined routes that store this information until it is reintroduced to deeper layers of the model, rather than discarding these tokens entirely. Further, we combine multiple routes and introduce an adapted auxiliary loss that accounts for all applied routes. Our method is not limited to the common transformer-based model - it can also be applied to state-space models. Unlike most current approaches, TREAD achieves this without architectural modifications. Finally, we show that our method reduces the computational cost and simultaneously boosts model performance on the standard benchmark ImageNet-1K 256 x 256 in class-conditional synthesis. Both of these benefits multiply to a convergence speedup of 9.55x at 400K training iterations compared to DiT and 25.39x compared to the best benchmark performance of DiT at 7M training iterations.
Authors: Andrew Bond, Jui-Hsien Wang, Long Mai, Erkut Erdem, Aykut Erdem
Abstract: Efficient neural representations for dynamic video scenes are critical for applications ranging from video compression to interactive simulations. Yet, existing methods often face challenges related to high memory usage, lengthy training times, and temporal consistency. To address these issues, we introduce a novel neural video representation that combines 3D Gaussian splatting with continuous camera motion modeling. By leveraging Neural ODEs, our approach learns smooth camera trajectories while maintaining an explicit 3D scene representation through Gaussians. Additionally, we introduce a spatiotemporal hierarchical learning strategy, progressively refining spatial and temporal features to enhance reconstruction quality and accelerate convergence. This memory-efficient approach achieves high-quality rendering at impressive speeds. Experimental results show that our hierarchical learning, combined with robust camera motion modeling, captures complex dynamic scenes with strong temporal consistency, achieving state-of-the-art performance across diverse video datasets in both high- and low-motion scenarios.
Authors: Srikar Yellapragada, Kowshik Thopalli, Vivek Narayanaswamy, Wesam Sakla, Yang Liu, Yamen Mubarka, Dimitris Samaras, Jayaraman J. Thiagarajan
Abstract: Vision Transformers (ViTs) have shown success across a variety of tasks due to their ability to capture global image representations. Recent studies have identified the existence of high-norm tokens in ViTs, which can interfere with unsupervised object discovery. To address this, the use of "registers" which are additional tokens that isolate high norm patch tokens while capturing global image-level information has been proposed. While registers have been studied extensively for object discovery, their generalization properties particularly in out-of-distribution (OOD) scenarios, remains underexplored. In this paper, we examine the utility of register token embeddings in providing additional features for improving generalization and anomaly rejection. To that end, we propose a simple method that combines the special CLS token embedding commonly employed in ViTs with the average-pooled register embeddings to create feature representations which are subsequently used for training a downstream classifier. We find that this enhances OOD generalization and anomaly rejection, while maintaining in-distribution (ID) performance. Extensive experiments across multiple ViT backbones trained with and without registers reveal consistent improvements of 2-4\% in top-1 OOD accuracy and a 2-3\% reduction in false positive rates for anomaly detection. Importantly, these gains are achieved without additional computational overhead.
Authors: Kaouther Messaoud, Matthieu Cord, Alexandre Alahi
Abstract: Existing vehicle trajectory prediction models struggle with generalizability, prediction uncertainties, and handling complex interactions. It is often due to limitations like complex architectures customized for a specific dataset and inefficient multimodal handling. We propose Perceiver with Register queries (PerReg+), a novel trajectory prediction framework that introduces: (1) Dual-Level Representation Learning via Self-Distillation (SD) and Masked Reconstruction (MR), capturing global context and fine-grained details. Additionally, our approach of reconstructing segmentlevel trajectories and lane segments from masked inputs with query drop, enables effective use of contextual information and improves generalization; (2) Enhanced Multimodality using register-based queries and pretraining, eliminating the need for clustering and suppression; and (3) Adaptive Prompt Tuning during fine-tuning, freezing the main architecture and optimizing a small number of prompts for efficient adaptation. PerReg+ sets a new state-of-the-art performance on nuScenes [1], Argoverse 2 [2], and Waymo Open Motion Dataset (WOMD) [3]. Remarkable, our pretrained model reduces the error by 6.8% on smaller datasets, and multi-dataset training enhances generalization. In cross-domain tests, PerReg+ reduces B-FDE by 11.8% compared to its non-pretrained variant.
Authors: Yachuan Li, Xavier Soria Poma, Yun Bai, Qian Xiao, Chaozhi Yang, Guanlin Li, Zongmin Li
Abstract: Transformer-based models have made significant progress in edge detection, but their high computational cost is prohibitive. Recently, vision Mamba have shown excellent ability in efficiently capturing long-range dependencies. Drawing inspiration from this, we propose a novel edge detector with Mamba, termed EDMB, to efficiently generate high-quality multi-granularity edges. In EDMB, Mamba is combined with a global-local architecture, therefore it can focus on both global information and fine-grained cues. The fine-grained cues play a crucial role in edge detection, but are usually ignored by ordinary Mamba. We design a novel decoder to construct learnable Gaussian distributions by fusing global features and fine-grained features. And the multi-grained edges are generated by sampling from the distributions. In order to make multi-granularity edges applicable to single-label data, we introduce Evidence Lower Bound loss to supervise the learning of the distributions. On the multi-label dataset BSDS500, our proposed EDMB achieves competitive single-granularity ODS 0.837 and multi-granularity ODS 0.851 without multi-scale test or extra PASCAL-VOC data. Remarkably, EDMB can be extended to single-label datasets such as NYUDv2 and BIPED. The source code is available at https://github.com/Li-yachuan/EDMB.
Authors: Hafiz Mughees Ahmad, Dario Morle, Afshin Rahimi
Abstract: Deep learning models have demonstrated remarkable performance across various computer vision tasks, yet their vulnerability to distribution shifts remains a critical challenge. Despite sophisticated neural network architectures, existing models often struggle to maintain consistent performance when confronted with Out-of-Distribution (OOD) samples, including natural corruptions, adversarial perturbations, and anomalous patterns. We introduce LayerMix, an innovative data augmentation approach that systematically enhances model robustness through structured fractal-based image synthesis. By meticulously integrating structural complexity into training datasets, our method generates semantically consistent synthetic samples that significantly improve neural network generalization capabilities. Unlike traditional augmentation techniques that rely on random transformations, LayerMix employs a structured mixing pipeline that preserves original image semantics while introducing controlled variability. Extensive experiments across multiple benchmark datasets, including CIFAR-10, CIFAR-100, ImageNet-200, and ImageNet-1K demonstrate LayerMixs superior performance in classification accuracy and substantially enhances critical Machine Learning (ML) safety metrics, including resilience to natural image corruptions, robustness against adversarial attacks, improved model calibration and enhanced prediction consistency. LayerMix represents a significant advancement toward developing more reliable and adaptable artificial intelligence systems by addressing the fundamental challenges of deep learning generalization. The code is available at https://github.com/ahmadmughees/layermix.
Authors: Alexander Valverde, Luis Solano
Abstract: In Costa Rica, an average of 5 tons of seashells are extracted from ecosystems annually. Confiscated seashells, cannot be returned to their ecosystems due to the lack of origin recognition. To address this issue, we developed a convolutional neural network (CNN) specifically for seashell identification. We built a dataset from scratch, consisting of approximately 19000 images from the Pacific and Caribbean coasts. Using this dataset, the model achieved a classification accuracy exceeding 85%. The model has been integrated into a user-friendly application, which has classified over 36,000 seashells to date, delivering real-time results within 3 seconds per image. To further enhance the system's accuracy, an anomaly detection mechanism was incorporated to filter out irrelevant or anomalous inputs, ensuring only valid seashell images are processed.
Authors: Christophe Lohou
Abstract: In this paper, we propose a topological classification of points for 2D discrete binary images. This classification is based on the values of the calculus of topological numbers. Six classes of points are proposed: isolated point, interior point, simple point, curve point, point of intersection of 3 curves, point of intersection of 4 curves. The number of configurations of each class is also given.
Authors: Afnan A. Shah
Abstract: Managing the massive annual gatherings of Hajj and Umrah presents significant challenges, particularly as the Saudi government aims to increase the number of pilgrims. Currently, around two million pilgrims attend Hajj and 26 million attend Umrah making crowd control especially in critical areas like the Grand Mosque during Tawaf, a major concern. Additional risks arise in managing dense crowds at key sites such as Arafat where the potential for stampedes, fires and pandemics poses serious threats to public safety. This research proposes a machine learning model to classify crowd density into three levels: moderate crowd, overcrowded and very dense crowd in video frames recorded during Hajj, with a flashing red light to alert organizers in real-time when a very dense crowd is detected. While current research efforts in processing Hajj surveillance videos focus solely on using CNN to detect abnormal behaviors, this research focuses more on high-risk crowds that can lead to disasters. Hazardous crowd conditions require a robust method, as incorrect classification could trigger unnecessary alerts and government intervention, while failure to classify could result in disaster. The proposed model integrates Local Binary Pattern (LBP) texture analysis, which enhances feature extraction for differentiating crowd density levels, along with edge density and area-based features. The model was tested on the KAU-Smart Crowd 'HAJJv2' dataset which contains 18 videos from various key locations during Hajj including 'Massaa', 'Jamarat', 'Arafat' and 'Tawaf'. The model achieved an accuracy rate of 87% with a 2.14% error percentage (misclassification rate), demonstrating its ability to detect and classify various crowd conditions effectively. That contributes to enhanced crowd management and safety during large-scale events like Hajj.
Authors: Golriz Hosseinimanesh, Farnoosh Ghadiri, Francois Guibault, Farida Cheriet, Julia Keren
Abstract: Designing a dental crown is a time-consuming and labor intensive process. Our goal is to simplify crown design and minimize the tediousness of making manual adjustments while still ensuring the highest level of accuracy and consistency. To this end, we present a new end- to-end deep learning approach, coined Dental Mesh Completion (DMC), to generate a crown mesh conditioned on a point cloud context. The dental context includes the tooth prepared to receive a crown and its surroundings, namely the two adjacent teeth and the three closest teeth in the opposing jaw. We formulate crown generation in terms of completing this point cloud context. A feature extractor first converts the input point cloud into a set of feature vectors that represent local regions in the point cloud. The set of feature vectors is then fed into a transformer to predict a new set of feature vectors for the missing region (crown). Subsequently, a point reconstruction head, followed by a multi-layer perceptron, is used to predict a dense set of points with normals. Finally, a differentiable point-to-mesh layer serves to reconstruct the crown surface mesh. We compare our DMC method to a graph-based convolutional neural network which learns to deform a crown mesh from a generic crown shape to the target geometry. Extensive experiments on our dataset demonstrate the effectiveness of our method, which attains an average of 0.062 Chamfer Distance.The code is available at:https://github.com/Golriz-code/DMC.gi
Authors: Xingang Li, Zhenghui Sha
Abstract: Computer-aided design (CAD) tools empower designers to design and modify 3D models through a series of CAD operations, commonly referred to as a CAD sequence. In scenarios where digital CAD files are not accessible, reverse engineering (RE) has been used to reconstruct 3D CAD models. Recent advances have seen the rise of data-driven approaches for RE, with a primary focus on converting 3D data, such as point clouds, into 3D models in boundary representation (B-rep) format. However, obtaining 3D data poses significant challenges, and B-rep models do not reveal knowledge about the 3D modeling process of designs. To this end, our research introduces a novel data-driven approach with an Image2CADSeq neural network model. This model aims to reverse engineer CAD models by processing images as input and generating CAD sequences. These sequences can then be translated into B-rep models using a solid modeling kernel. Unlike B-rep models, CAD sequences offer enhanced flexibility to modify individual steps of model creation, providing a deeper understanding of the construction process of CAD models. To quantitatively and rigorously evaluate the predictive performance of the Image2CADSeq model, we have developed a multi-level evaluation framework for model assessment. The model was trained on a specially synthesized dataset, and various network architectures were explored to optimize the performance. The experimental and validation results show great potential for the model in generating CAD sequences from 2D image data.
Authors: Zhenghui Zhao, Chen Wu, Lixiang Ru, Di Wang, Hongruixuan Chen, Cuiqun Chen
Abstract: Existing Weakly-Supervised Change Detection (WSCD) methods often encounter the problem of "instance lumping" under scene-level supervision, particularly in scenarios with a dense distribution of changed instances (i.e., changed objects). In these scenarios, unchanged pixels between changed instances are also mistakenly identified as changed, causing multiple changes to be mistakenly viewed as one. In practical applications, this issue prevents the accurate quantification of the number of changes. To address this issue, we propose a Dense Instance Separation (DISep) method as a plug-and-play solution, refining pixel features from a unified instance perspective under scene-level supervision. Specifically, our DISep comprises a three-step iterative training process: 1) Instance Localization: We locate instance candidate regions for changed pixels using high-pass class activation maps. 2) Instance Retrieval: We identify and group these changed pixels into different instance IDs through connectivity searching. Then, based on the assigned instance IDs, we extract corresponding pixel-level features on a per-instance basis. 3) Instance Separation: We introduce a separation loss to enforce intra-instance pixel consistency in the embedding space, thereby ensuring separable instance feature representations. The proposed DISep adds only minimal training cost and no inference cost. It can be seamlessly integrated to enhance existing WSCD methods. We achieve state-of-the-art performance by enhancing {three Transformer-based and four ConvNet-based methods} on the LEVIR-CD, WHU-CD, DSIFN-CD, SYSU-CD, and CDD datasets. Additionally, our DISep can be used to improve fully-supervised change detection methods. Code is available at https://github.com/zhenghuizhao/Plug-and-Play-DISep-for-Change-Detection.
URLs: https://github.com/zhenghuizhao/Plug-and-Play-DISep-for-Change-Detection.
Authors: Sun-Hyuk Choi, Hayoung Jo, Seong-Whan Lee
Abstract: Referring video object segmentation aims to segment objects within a video corresponding to a given text description. Existing transformer-based temporal modeling approaches face challenges related to query inconsistency and the limited consideration of context. Query inconsistency produces unstable masks of different objects in the middle of the video. The limited consideration of context leads to the segmentation of incorrect objects by failing to adequately account for the relationship between the given text and instances. To address these issues, we propose the Multi-context Temporal Consistency Module (MTCM), which consists of an Aligner and a Multi-Context Enhancer (MCE). The Aligner removes noise from queries and aligns them to achieve query consistency. The MCE predicts text-relevant queries by considering multi-context. We applied MTCM to four different models, increasing performance across all of them, particularly achieving 47.6 J&F on the MeViS. Code is available at https://github.com/Choi58/MTCM.
Authors: Yapeng Li, Yong Luo, Lefei Zhang, Zengmao Wang, Bo Du
Abstract: Transformer has been extensively explored for hyperspectral image (HSI) classification. However, transformer poses challenges in terms of speed and memory usage because of its quadratic computational complexity. Recently, the Mamba model has emerged as a promising approach, which has strong long-distance modeling capabilities while maintaining a linear computational complexity. However, representing the HSI is challenging for the Mamba due to the requirement for an integrated spatial and spectral understanding. To remedy these drawbacks, we propose a novel HSI classification model based on a Mamba model, named MambaHSI, which can simultaneously model long-range interaction of the whole image and integrate spatial and spectral information in an adaptive manner. Specifically, we design a spatial Mamba block (SpaMB) to model the long-range interaction of the whole image at the pixel-level. Then, we propose a spectral Mamba block (SpeMB) to split the spectral vector into multiple groups, mine the relations across different spectral groups, and extract spectral features. Finally, we propose a spatial-spectral fusion module (SSFM) to adaptively integrate spatial and spectral features of a HSI. To our best knowledge, this is the first image-level HSI classification model based on the Mamba. We conduct extensive experiments on four diverse HSI datasets. The results demonstrate the effectiveness and superiority of the proposed model for HSI classification. This reveals the great potential of Mamba to be the next-generation backbone for HSI models. Codes are available at https://github.com/li-yapeng/MambaHSI .
Authors: Yifan Xu, Vineet Kamat, Carol Menassa
Abstract: In assistive robotics serving people with disabilities (PWD), accurate place recognition in built environments is crucial to ensure that robots navigate and interact safely within diverse indoor spaces. Language interfaces, particularly those powered by Large Language Models (LLM) and Vision Language Models (VLM), hold significant promise in this context, as they can interpret visual scenes and correlate them with semantic information. However, such interfaces are also known for their hallucinated predictions. In addition, language instructions provided by humans can also be ambiguous and lack precise details about specific locations, objects, or actions, exacerbating the hallucination issue. In this work, we introduce Seeing with Partial Certainty (SwPC) - a framework designed to measure and align uncertainty in VLM-based place recognition, enabling the model to recognize when it lacks confidence and seek assistance when necessary. This framework is built on the theory of conformal prediction to provide statistical guarantees on place recognition while minimizing requests for human help in complex indoor environment settings. Through experiments on the widely used richly-annotated scene dataset Matterport3D, we show that SwPC significantly increases the success rate and decreases the amount of human intervention required relative to the prior art. SwPC can be utilized with any VLMs directly without requiring model fine-tuning, offering a promising, lightweight approach to uncertainty modeling that complements and scales alongside the expanding capabilities of foundational models.
Authors: Hojun Lim, Heecheol Yoo, Jinwoo Lee, Seungmin Jeon, Hyeongseok Jeon
Abstract: Deep neural network (DNN) based perception models are indispensable in the development of autonomous vehicles (AVs). However, their reliance on large-scale, high-quality data is broadly recognized as a burdensome necessity due to the substantial cost of data acquisition and labeling. Further, the issue is not a one-time concern, as AVs might need a new dataset if they are to be deployed to another region (real-target domain) that the in-hand dataset within the real-source domain cannot incorporate. To mitigate this burden, we propose leveraging synthetic environments as an auxiliary domain where the characteristics of real domains are reproduced. This approach could enable indirect experience about the real-target domain in a time- and cost-effective manner. As a practical demonstration of our methodology, nuScenes and South Korea are employed to represent real-source and real-target domains, respectively. That means we construct digital twins for several regions of South Korea, and the data-acquisition framework of nuScenes is reproduced. Blending the aforementioned components within a simulator allows us to obtain a synthetic-fusion domain in which we forge our novel driving dataset, MORDA: Mixture Of Real-domain characteristics for synthetic-data-assisted Domain Adaptation. To verify the value of synthetic features that MORDA provides in learning about driving environments of South Korea, 2D/3D detectors are trained solely on a combination of nuScenes and MORDA. Afterward, their performance is evaluated on the unforeseen real-world dataset (AI-Hub) collected in South Korea. Our experiments present that MORDA can significantly improve mean Average Precision (mAP) on AI-Hub dataset while that on nuScenes is retained or slightly enhanced.
Authors: Lei Li, Xinglin Zhang, Jun Liang, Tao Chen
Abstract: Deep learning models in medical imaging face dual challenges: domain shift, where models perform poorly when deployed in settings different from their training environment, and class imbalance, where certain disease conditions are naturally underrepresented. We present Imbalance-Aware Domain Adaptation (IADA), a novel framework that simultaneously tackles both challenges through three key components: (1) adaptive feature learning with class-specific attention mechanisms, (2) balanced domain alignment with dynamic weighting, and (3) adaptive threshold optimization. Our theoretical analysis establishes convergence guarantees and complexity bounds. Through extensive experiments on embryo development assessment across four imaging modalities, IADA demonstrates significant improvements over existing methods, achieving up to 25.19\% higher accuracy while maintaining balanced performance across classes. In challenging scenarios with low-quality imaging systems, IADA shows robust generalization with AUC improvements of up to 12.56\%. These results demonstrate IADA's potential for developing reliable and equitable medical imaging systems for diverse clinical settings. The code is made public available at \url{https://github.com/yinghemedical/imbalance-aware_domain_adaptation}
URLs: https://github.com/yinghemedical/imbalance-aware_domain_adaptation
Authors: Yi Lin, Lin Gu, Ziteng Cui, Shenghan Su, Yumo Hao, Yingtao Tian, Tatsuya Harada, Jianfei Yang
Abstract: From Paleolithic cave paintings to Impressionism, human painting has evolved to depict increasingly complex and detailed scenes, conveying more nuanced messages. This paper attempts to emerge this artistic capability by simulating the evolutionary pressures that enhance visual communication efficiency. Specifically, we present a model with a stroke branch and a palette branch that together simulate human-like painting. The palette branch learns a limited colour palette, while the stroke branch parameterises each stroke using B\'ezier curves to render an image, subsequently evaluated by a high-level recognition module. We quantify the efficiency of visual communication by measuring the recognition accuracy achieved with machine vision. The model then optimises the control points and colour choices for each stroke to maximise recognition accuracy with minimal strokes and colours. Experimental results show that our model achieves superior performance in high-level recognition tasks, delivering artistic expression and aesthetic appeal, especially in abstract sketches. Additionally, our approach shows promise as an efficient bit-level image compression technique, outperforming traditional methods.
Authors: Hangzhou He, Lei Zhu, Xinliang Zhang, Shuang Zeng, Qian Chen, Yanye Lu
Abstract: Concept Bottleneck Models (CBMs) offer inherent interpretability by initially translating images into human-comprehensible concepts, followed by a linear combination of these concepts for classification. However, the annotation of concepts for visual recognition tasks requires extensive expert knowledge and labor, constraining the broad adoption of CBMs. Recent approaches have leveraged the knowledge of large language models to construct concept bottlenecks, with multimodal models like CLIP subsequently mapping image features into the concept feature space for classification. Despite this, the concepts produced by language models can be verbose and may introduce non-visual attributes, which hurts accuracy and interpretability. In this study, we investigate to avoid these issues by constructing CBMs directly from multimodal models. To this end, we adopt common words as base concept vocabulary and leverage auxiliary unlabeled images to construct a Vision-to-Concept (V2C) tokenizer that can explicitly quantize images into their most relevant visual concepts, thus creating a vision-oriented concept bottleneck tightly coupled with the multimodal model. This leads to our V2C-CBM which is training efficient and interpretable with high accuracy. Our V2C-CBM has matched or outperformed LLM-supervised CBMs on various visual classification benchmarks, validating the efficacy of our approach.
Authors: Qi Chen, Changli Wu, Jiayi Ji, Yiwei Ma, Danni Yang, Xiaoshuai Sun
Abstract: 3D Referring Expression Segmentation (3D-RES) aims to segment point cloud scenes based on a given expression. However, existing 3D-RES approaches face two major challenges: feature ambiguity and intent ambiguity. Feature ambiguity arises from information loss or distortion during point cloud acquisition due to limitations such as lighting and viewpoint. Intent ambiguity refers to the model's equal treatment of all queries during the decoding process, lacking top-down task-specific guidance. In this paper, we introduce an Image enhanced Prompt Decoding Network (IPDN), which leverages multi-view images and task-driven information to enhance the model's reasoning capabilities. To address feature ambiguity, we propose the Multi-view Semantic Embedding (MSE) module, which injects multi-view 2D image information into the 3D scene and compensates for potential spatial information loss. To tackle intent ambiguity, we designed a Prompt-Aware Decoder (PAD) that guides the decoding process by deriving task-driven signals from the interaction between the expression and visual features. Comprehensive experiments demonstrate that IPDN outperforms the state-ofthe-art by 1.9 and 4.2 points in mIoU metrics on the 3D-RES and 3D-GRES tasks, respectively.
Authors: Ziyang Gao, Yong Tian, Shih-Chi Lin, Junghua Lin
Abstract: In the medical field, accurate diagnosis of lung cancer is crucial for treatment. Traditional manual analysis methods have significant limitations in terms of accuracy and efficiency. To address this issue, this paper proposes a deep learning network framework based on the pre-trained MobileNetV2 model, initialized with weights from the ImageNet-1K dataset (version 2). The last layer of the model (the fully connected layer) is replaced with a new fully connected layer, and a softmax activation function is added to efficiently classify three types of lung cancer CT scan images. Experimental results show that the model achieves an accuracy of 99.6% on the test set, with significant improvements in feature extraction compared to traditional models.With the rapid development of artificial intelligence technologies, deep learning applications in medical image processing are bringing revolutionary changes to the healthcare industry. AI-based lung cancer detection systems can significantly improve diagnostic efficiency, reduce the workload of doctors, and occupy an important position in the global healthcare market. The potential of AI to improve diagnostic accuracy, reduce medical costs, and promote precision medicine will have a profound impact on the future development of the healthcare industry.
Authors: Xiaojie Li, Yibo Yang, Jianlong Wu, David A. Clifton, Yue Yu, Bernard Ghanem, Min Zhang
Abstract: Few-shot class-incremental learning (FSCIL) involves learning new classes from limited data while retaining prior knowledge, and often results in catastrophic forgetting. Existing methods either freeze backbone networks to preserve knowledge, which limits adaptability, or rely on additional modules or prompts, introducing inference overhead. To this end, we propose Continuous Knowledge-Preserving Decomposition for FSCIL (CKPD-FSCIL), a framework that decomposes a model's weights into two parts: one that compacts existing knowledge (knowledge-sensitive components) and another that carries redundant capacity to accommodate new abilities (redundant-capacity components). The decomposition is guided by a covariance matrix from replay samples, ensuring principal components align with classification abilities. During adaptation, we freeze the knowledge-sensitive components and only adapt the redundant-capacity components, fostering plasticity while minimizing interference without changing the architecture or increasing overhead. Additionally, CKPD introduces an adaptive layer selection strategy to identify layers with redundant capacity, dynamically allocating adapters. Experiments on multiple benchmarks show that CKPD-FSCIL outperforms state-of-the-art methods.
Authors: Yingjie Chen, Yifang Men, Yuan Yao, Miaomiao Cui, Liefeng Bo
Abstract: Motion-controllable image animation is a fundamental task with a wide range of potential applications. Recent works have made progress in controlling camera or object motion via various motion representations, while they still struggle to support collaborative camera and object motion control with adaptive control granularity. To this end, we introduce 3D-aware motion representation and propose an image animation framework, called Perception-as-Control, to achieve fine-grained collaborative motion control. Specifically, we construct 3D-aware motion representation from a reference image, manipulate it based on interpreted user intentions, and perceive it from different viewpoints. In this way, camera and object motions are transformed into intuitive, consistent visual changes. Then, the proposed framework leverages the perception results as motion control signals, enabling it to support various motion-related video synthesis tasks in a unified and flexible way. Experiments demonstrate the superiority of the proposed framework. For more details and qualitative results, please refer to our project webpage: https://chen-yingjie.github.io/projects/Perception-as-Control.
URLs: https://chen-yingjie.github.io/projects/Perception-as-Control.
Authors: Ronghao Dang, Yuqian Yuan, Wenqi Zhang, Yifei Xin, Boqiang Zhang, Long Li, Liuyi Wang, Qinyang Zeng, Xin Li, Lidong Bing
Abstract: The enhancement of generalization in robots by large vision-language models (LVLMs) is increasingly evident. Therefore, the embodied cognitive abilities of LVLMs based on egocentric videos are of great interest. However, current datasets for embodied video question answering lack comprehensive and systematic evaluation frameworks. Critical embodied cognitive issues, such as robotic self-cognition, dynamic scene perception, and hallucination, are rarely addressed. To tackle these challenges, we propose ECBench, a high-quality benchmark designed to systematically evaluate the embodied cognitive abilities of LVLMs. ECBench features a diverse range of scene video sources, open and varied question formats, and 30 dimensions of embodied cognition. To ensure quality, balance, and high visual dependence, ECBench uses class-independent meticulous human annotation and multi-round question screening strategies. Additionally, we introduce ECEval, a comprehensive evaluation system that ensures the fairness and rationality of the indicators. Utilizing ECBench, we conduct extensive evaluations of proprietary, open-source, and task-specific LVLMs. ECBench is pivotal in advancing the embodied cognitive capabilities of LVLMs, laying a solid foundation for developing reliable core models for embodied agents. All data and code are available at https://github.com/Rh-Dang/ECBench.
Authors: Laurenz Ruzicka, Alexander Spenke, Stephan Bergmann, Gerd Nolden, Bernhard Kohn, Clemens Heitzinger
Abstract: Fingerprint mosaicking, which is the process of combining multiple fingerprint images into a single master fingerprint, is an essential process in modern biometric systems. However, it is prone to errors that can significantly degrade fingerprint image quality. This paper proposes a novel deep learning-based approach to detect and score mosaicking artifacts in fingerprint images. Our method leverages a self-supervised learning framework to train a model on large-scale unlabeled fingerprint data, eliminating the need for manual artifact annotation. The proposed model effectively identifies mosaicking errors, achieving high accuracy on various fingerprint modalities, including contactless, rolled, and pressed fingerprints and furthermore proves to be robust to different data sources. Additionally, we introduce a novel mosaicking artifact score to quantify the severity of errors, enabling automated evaluation of fingerprint images. By addressing the challenges of mosaicking artifact detection, our work contributes to improving the accuracy and reliability of fingerprint-based biometric systems.
Authors: Rujie Wu, Xiaojian Ma, Hai Ci, Yue Fan, Yuxuan Wang, Haozhe Zhao, Qing Li, Yizhou Wang
Abstract: This paper introduce LongViTU, a large-scale (~121k QA pairs, ~900h videos), automatically generated dataset for long-form video understanding. We developed a systematic approach that organizes videos into a hierarchical tree structure and incorporates self-revision mechanisms to ensure high-quality QA pairs. Each QA pair in LongViTU features: 1) long-term context (average certificate length of 4.6 minutes); 2) rich knowledge and condensed reasoning (commonsense, causality, planning, etc.); and 3) explicit timestamp labels for relevant events. LongViTU also serves as a benchmark for instruction following in long-form and streaming video understanding. We evaluate the open-source state-of-the-art long video understanding model, LongVU, and the commercial model, Gemini-1.5-Pro, on our benchmark. They achieve GPT-4 scores of 49.9 and 52.3, respectively, underscoring the substantial challenge posed by our benchmark. Further supervised fine-tuning (SFT) on LongVU led to performance improvements of 12.0% on our benchmark, 2.2% on the in-distribution (ID) benchmark EgoSchema, 1.0%, 2.2% and 1.2% on the out-of-distribution (OOD) benchmarks VideoMME (Long), WorldQA and OpenEQA, respectively. These outcomes demonstrate LongViTU's high data quality and robust OOD generalizability.
Authors: Hao Wen, Ziqian Lu, Fengli Shen, Zhe-Ming Lu, Jialin Cui
Abstract: Human skeleton information is important in skeleton-based action recognition, which provides a simple and efficient way to describe human pose. However, existing skeleton-based methods focus more on the skeleton, ignoring the objects interacting with humans, resulting in poor performance in recognizing actions that involve object interactions. We propose a new action recognition framework introducing object nodes to supplement absent interactive object information. We also propose Spatial Temporal Variable Graph Convolutional Networks (ST-VGCN) to effectively model the Variable Graph (VG) containing object nodes. Specifically, in order to validate the role of interactive object information, by leveraging a simple self-training approach, we establish a new dataset, JXGC 24, and an extended dataset, NTU RGB+D+Object 60, including more than 2 million additional object nodes. At the same time, we designe the Variable Graph construction method to accommodate a variable number of nodes for graph structure. Additionally, we are the first to explore the overfitting issue introduced by incorporating additional object information, and we propose a VG-based data augmentation method to address this issue, called Random Node Attack. Finally, regarding the network structure, we introduce two fusion modules, CAF and WNPool, along with a novel Node Balance Loss, to enhance the comprehensive performance by effectively fusing and balancing skeleton and object node information. Our method surpasses the previous state-of-the-art on multiple skeleton-based action recognition benchmarks. The accuracy of our method on NTU RGB+D 60 cross-subject split is 96.7\%, and on cross-view split, it is 99.2\%.
Authors: Jiaxing Zhao, Boyuan Sun, Xiang Chen, Xihan Wei, Qibin Hou
Abstract: In this paper, we introduce LLaVA-Octopus, a novel video multimodal large language model. LLaVA-Octopus adaptively weights features from different visual projectors based on user instructions, enabling us to leverage the complementary strengths of each projector. We observe that different visual projectors exhibit distinct characteristics when handling specific tasks. For instance, some projectors excel at capturing static details, while others are more effective at processing temporal information, and some are better suited for tasks requiring temporal coherence. By dynamically adjusting feature weights according to user instructions, LLaVA-Octopus dynamically selects and combines the most suitable features, significantly enhancing the model's performance in multimodal tasks. Experimental results demonstrate that LLaVA-Octopus achieves excellent performance across multiple benchmarks, especially in tasks such as multimodal understanding, visual question answering, and video understanding, highlighting its broad application potential.
Authors: Huabin Liu, Filip Ilievski, Cees G. M. Snoek
Abstract: This paper proposes the first video-grounded entailment tree reasoning method for commonsense video question answering (VQA). Despite the remarkable progress of large visual-language models (VLMs), there are growing concerns that they learn spurious correlations between videos and likely answers, reinforced by their black-box nature and remaining benchmarking biases. Our method explicitly grounds VQA tasks to video fragments in four steps: entailment tree construction, video-language entailment verification, tree reasoning, and dynamic tree expansion. A vital benefit of the method is its generalizability to current video and image-based VLMs across reasoning types. To support fair evaluation, we devise a de-biasing procedure based on large-language models that rewrites VQA benchmark answer sets to enforce model reasoning. Systematic experiments on existing and de-biased benchmarks highlight the impact of our method components across benchmarks, VLMs, and reasoning types.
Authors: Laurenz Ruzicka, Bernhard Kohn, Clemens Heitzinger
Abstract: Contactless fingerprint recognition systems offer a hygienic, user-friendly, and efficient alternative to traditional contact-based methods. However, their accuracy heavily relies on precise fingertip detection and segmentation, particularly under challenging background conditions. This paper introduces TipSegNet, a novel deep learning model that achieves state-of-the-art performance in segmenting fingertips directly from grayscale hand images. TipSegNet leverages a ResNeXt-101 backbone for robust feature extraction, combined with a Feature Pyramid Network (FPN) for multi-scale representation, enabling accurate segmentation across varying finger poses and image qualities. Furthermore, we employ an extensive data augmentation strategy to enhance the model's generalizability and robustness. TipSegNet outperforms existing methods, achieving a mean Intersection over Union (mIoU) of 0.987 and an accuracy of 0.999, representing a significant advancement in contactless fingerprint segmentation. This enhanced accuracy has the potential to substantially improve the reliability and effectiveness of contactless biometric systems in real-world applications.
Authors: Yoseob Han, Dufan Wu, Kyungsang Kim, Quanzheng Li
Abstract: Objective: There exist several X-ray computed tomography (CT) scanning strategies to reduce a radiation dose, such as (1) sparse-view CT, (2) low-dose CT, and (3) region-of-interest (ROI) CT (called interior tomography). To further reduce the dose, the sparse-view and/or low-dose CT settings can be applied together with interior tomography. Interior tomography has various advantages in terms of reducing the number of detectors and decreasing the X-ray radiation dose. However, a large patient or small field-of-view (FOV) detector can cause truncated projections, and then the reconstructed images suffer from severe cupping artifacts. In addition, although the low-dose CT can reduce the radiation exposure dose, analytic reconstruction algorithms produce image noise. Recently, many researchers have utilized image-domain deep learning (DL) approaches to remove each artifact and demonstrated impressive performances, and the theory of deep convolutional framelets supports the reason for the performance improvement. Approach: In this paper, we found that the image-domain convolutional neural network (CNN) is difficult to solve coupled artifacts, based on deep convolutional framelets. Significance: To address the coupled problem, we decouple it into two sub-problems: (i) image domain noise reduction inside truncated projection to solve low-dose CT problem and (ii) extrapolation of projection outside truncated projection to solve the ROI CT problem. The decoupled sub-problems are solved directly with a novel proposed end-to-end learning using dual-domain CNNs. Main results: We demonstrate that the proposed method outperforms the conventional image-domain deep learning methods, and a projection-domain CNN shows better performance than the image-domain CNNs which are commonly used by many researchers.
Authors: Shiqi Cao, Liangjian Deng, Shangqi Deng
Abstract: The implementation of diffusion-based pansharpening task is predominantly constrained by its slow inference speed, which results from numerous sampling steps. Despite the existing techniques aiming to accelerate sampling, they often compromise performance when fusing multi-source images. To ease this limitation, we introduce a novel and efficient diffusion model named Diffusion Model for Pansharpening by Inferring Residual Inference (ResPanDiff), which significantly reduces the number of diffusion steps without sacrificing the performance to tackle pansharpening task. In ResPanDiff, we innovatively propose a Markov chain that transits from noisy residuals to the residuals between the LRMS and HRMS images, thereby reducing the number of sampling steps and enhancing performance. Additionally, we design the latent space to help model extract more features at the encoding stage, Shallow Cond-Injection~(SC-I) to help model fetch cond-injected hidden features with higher dimensions, and loss functions to give a better guidance for the residual generation task. enabling the model to achieve superior performance in residual generation. Furthermore, experimental evaluations on pansharpening datasets demonstrate that the proposed method achieves superior outcomes compared to recent state-of-the-art~(SOTA) techniques, requiring only 15 sampling steps, which reduces over $90\%$ step compared with the benchmark diffusion models. Our experiments also include thorough discussions and ablation studies to underscore the effectiveness of our approach.
Authors: Haoyi Xiu, Xin Liu, Taehoon Kim, Kyoung-Sook Kim
Abstract: The pre-training and fine-tuning paradigm has revolutionized satellite remote sensing applications. However, this approach remains largely underexplored for airborne laser scanning (ALS), an important technology for applications such as forest management and urban planning. In this study, we address this gap by constructing a large-scale ALS point cloud dataset and evaluating its impact on downstream applications. Our dataset comprises ALS point clouds collected across the contiguous United States, provided by the United States Geological Survey's 3D Elevation Program. To ensure efficient data collection while capturing diverse land cover and terrain types, we introduce a geospatial sampling method that selects point cloud tiles based on land cover maps and digital elevation models. As a baseline self-supervised learning model, we adopt BEV-MAE, a state-of-the-art masked autoencoder for 3D outdoor point clouds, and pre-train it on the constructed dataset. The pre-trained models are subsequently fine-tuned for downstream tasks, including tree species classification, terrain scene recognition, and point cloud semantic segmentation. Our results show that the pre-trained models significantly outperform their scratch counterparts across all downstream tasks, demonstrating the transferability of the representations learned from the proposed dataset. Furthermore, we observe that scaling the dataset using our geospatial sampling method consistently enhances performance, whereas pre-training on datasets constructed with random sampling fails to achieve similar improvements. These findings highlight the utility of the constructed dataset and the effectiveness of our sampling strategy in the pre-training and fine-tuning paradigm. The source code and pre-trained models will be made publicly available at \url{https://github.com/martianxiu/ALS_pretraining}.
Authors: Van Thien Nguyen, William Guicquero, Gilles Sicard
Abstract: Even if Application-Specific Integrated Circuits (ASIC) have proven to be a relevant choice for integrating inference at the edge, they are often limited in terms of applicability. In this paper, we demonstrate that an ASIC neural network accelerator dedicated to image processing can be applied to multiple tasks of different levels: image classification and compression, while requiring a very limited hardware. The key component is a reconfigurable, mixed-precision (3b/2b/1b) encoder that takes advantage of proper weight and activation quantizations combined with convolutional layer structural pruning to lower hardware-related constraints (memory and computing). We introduce an automatic adaptation of linear symmetric quantizer scaling factors to perform quantized levels equalization, aiming at stabilizing quinary and ternary weights training. In addition, a proposed layer-shared Bit-Shift Normalization significantly simplifies the implementation of the hardware-expensive Batch Normalization. For a specific configuration in which the encoder design only requires 1Mb, the classification accuracy reaches 87.5% on CIFAR-10. Besides, we also show that this quantized encoder can be used to compress image patch-by-patch while the reconstruction can performed remotely, by a dedicated full-frame decoder. This solution typically enables an end-to-end compression almost without any block artifacts, outperforming patch-based state-of-the-art techniques employing a patch-constant bitrate.
Authors: Yuhong Zhang, Jing Lin, Ailing Zeng, Guanlin Wu, Shunlin Lu, Yurong Fu, Yuanhao Cai, Ruimao Zhang, Haoqian Wang, Lei Zhang
Abstract: In this paper, we introduce Motion-X++, a large-scale multimodal 3D expressive whole-body human motion dataset. Existing motion datasets predominantly capture body-only poses, lacking facial expressions, hand gestures, and fine-grained pose descriptions, and are typically limited to lab settings with manually labeled text descriptions, thereby restricting their scalability. To address this issue, we develop a scalable annotation pipeline that can automatically capture 3D whole-body human motion and comprehensive textural labels from RGB videos and build the Motion-X dataset comprising 81.1K text-motion pairs. Furthermore, we extend Motion-X into Motion-X++ by improving the annotation pipeline, introducing more data modalities, and scaling up the data quantities. Motion-X++ provides 19.5M 3D whole-body pose annotations covering 120.5K motion sequences from massive scenes, 80.8K RGB videos, 45.3K audios, 19.5M frame-level whole-body pose descriptions, and 120.5K sequence-level semantic labels. Comprehensive experiments validate the accuracy of our annotation pipeline and highlight Motion-X++'s significant benefits for generating expressive, precise, and natural motion with paired multimodal labels supporting several downstream tasks, including text-driven whole-body motion generation,audio-driven motion generation, 3D whole-body human mesh recovery, and 2D whole-body keypoints estimation, etc.
Authors: Naval Kishore Mehta, Arvind, Shyam Sunder Prasad, Sumeet Saurav, Sanjay Singh
Abstract: Monitoring complex assembly processes is critical for maintaining productivity and ensuring compliance with assembly standards. However, variability in human actions and subjective task preferences complicate accurate task anticipation and guidance. To address these challenges, we introduce the Multi-Modal Transformer Fusion and Recurrent Units (MMTFRU) Network for egocentric activity anticipation, utilizing multimodal fusion to improve prediction accuracy. Integrated with the Operator Action Monitoring Unit (OAMU), the system provides proactive operator guidance, preventing deviations in the assembly process. OAMU employs two strategies: (1) Top-5 MMTF-RU predictions, combined with a reference graph and an action dictionary, for next-step recommendations; and (2) Top-1 MMTF-RU predictions, integrated with a reference graph, for detecting sequence deviations and predicting anomaly scores via an entropy-informed confidence mechanism. We also introduce Time-Weighted Sequence Accuracy (TWSA) to evaluate operator efficiency and ensure timely task completion. Our approach is validated on the industrial Meccano dataset and the largescale EPIC-Kitchens-55 dataset, demonstrating its effectiveness in dynamic environments.
Authors: Dewei Zhou, Ji Xie, Zongxin Yang, Yi Yang
Abstract: The growing demand for controllable outputs in text-to-image generation has driven significant advancements in multi-instance generation (MIG), enabling users to define both instance layouts and attributes. Currently, the state-of-the-art methods in MIG are primarily adapter-based. However, these methods necessitate retraining a new adapter each time a more advanced model is released, resulting in significant resource consumption. A methodology named Depth-Driven Decoupled Instance Synthesis (3DIS) has been introduced, which decouples MIG into two distinct phases: 1) depth-based scene construction and 2) detail rendering with widely pre-trained depth control models. The 3DIS method requires adapter training solely during the scene construction phase, while enabling various models to perform training-free detail rendering. Initially, 3DIS focused on rendering techniques utilizing U-Net architectures such as SD1.5, SD2, and SDXL, without exploring the potential of recent DiT-based models like FLUX. In this paper, we present 3DIS-FLUX, an extension of the 3DIS framework that integrates the FLUX model for enhanced rendering capabilities. Specifically, we employ the FLUX.1-Depth-dev model for depth map controlled image generation and introduce a detail renderer that manipulates the Attention Mask in FLUX's Joint Attention mechanism based on layout information. This approach allows for the precise rendering of fine-grained attributes of each instance. Our experimental results indicate that 3DIS-FLUX, leveraging the FLUX model, outperforms the original 3DIS method, which utilized SD2 and SDXL, and surpasses current state-of-the-art adapter-based methods in terms of both performance and image quality. Project Page: https://limuloo.github.io/3DIS/.
Authors: Xiang Zhang, Chenchen Fu, Yufei Cui, Lan Yi, Yuyang Sun, Weiwei Wu, Xue Liu
Abstract: Real-time object detection takes an essential part in the decision-making process of numerous real-world applications, including collision avoidance and path planning in autonomous driving systems. This paper presents a novel real-time streaming perception method named CorrDiff, designed to tackle the challenge of delays in real-time detection systems. The main contribution of CorrDiff lies in its adaptive delay-aware detector, which is able to utilize runtime-estimated temporal cues to predict objects' locations for multiple future frames, and selectively produce predictions that matches real-world time, effectively compensating for any communication and computational delays. The proposed model outperforms current state-of-the-art methods by leveraging motion estimation and feature enhancement, both for 1) single-frame detection for the current frame or the next frame, in terms of the metric mAP, and 2) the prediction for (multiple) future frame(s), in terms of the metric sAP (The sAP metric is to evaluate object detection algorithms in streaming scenarios, factoring in both latency and accuracy). It demonstrates robust performance across a range of devices, from powerful Tesla V100 to modest RTX 2080Ti, achieving the highest level of perceptual accuracy on all platforms. Unlike most state-of-the-art methods that struggle to complete computation within a single frame on less powerful devices, CorrDiff meets the stringent real-time processing requirements on all kinds of devices. The experimental results emphasize the system's adaptability and its potential to significantly improve the safety and reliability for many real-world systems, such as autonomous driving. Our code is completely open-sourced and is available at https://anonymous.4open.science/r/CorrDiff.
Authors: Ali Rohan, Md Junayed Hasan, Andrei Petrovski
Abstract: Depth estimation (DE) provides spatial information about a scene and enables tasks such as 3D reconstruction, object detection, and scene understanding. Recently, there has been an increasing interest in using deep learning (DL)-based methods for DE. Traditional techniques rely on handcrafted features that often struggle to generalise to diverse scenes and require extensive manual tuning. However, DL models for DE can automatically extract relevant features from input data, adapt to various scene conditions, and generalise well to unseen environments. Numerous DL-based methods have been developed, making it necessary to survey and synthesize the state-of-the-art (SOTA). Previous reviews on DE have mainly focused on either monocular or stereo-based techniques, rather than comprehensively reviewing DE. Furthermore, to the best of our knowledge, there is no systematic literature review (SLR) that comprehensively focuses on DE. Therefore, this SLR study is being conducted. Initially, electronic databases were searched for relevant publications, resulting in 1284 publications. Using defined exclusion and quality criteria, 128 publications were shortlisted and further filtered to select 59 high-quality primary studies. These studies were analysed to extract data and answer defined research questions. Based on the results, DL methods were developed for mainly three different types of DE: monocular, stereo, and multi-view. 20 publicly available datasets were used to train, test, and evaluate DL models for DE, with KITTI, NYU Depth V2, and Make 3D being the most used datasets. 29 evaluation metrics were used to assess the performance of DE. 35 base models were reported in the primary studies, and the top five most-used base models were ResNet-50, ResNet-18, ResNet-101, U-Net, and VGG-16. Finally, the lack of ground truth data was among the most significant challenges reported by primary studies.
Authors: Siyu Liu, Zheng-Peng Duan, Jia OuYang, Jiayi Fu, Hyunhee Park, Zikun Liu, Chun-Le Guo, Chongyi Li
Abstract: Blind face restoration is a highly ill-posed problem due to the lack of necessary context. Although existing methods produce high-quality outputs, they often fail to faithfully preserve the individual's identity. In this paper, we propose a personalized face restoration method, FaceMe, based on a diffusion model. Given a single or a few reference images, we use an identity encoder to extract identity-related features, which serve as prompts to guide the diffusion model in restoring high-quality and identity-consistent facial images. By simply combining identity-related features, we effectively minimize the impact of identity-irrelevant features during training and support any number of reference image inputs during inference. Additionally, thanks to the robustness of the identity encoder, synthesized images can be used as reference images during training, and identity changing during inference does not require fine-tuning the model. We also propose a pipeline for constructing a reference image training pool that simulates the poses and expressions that may appear in real-world scenarios. Experimental results demonstrate that our FaceMe can restore high-quality facial images while maintaining identity consistency, achieving excellent performance and robustness.
Authors: Xuyang Liu, Ziming Wang, Yuhang Han, Yingyao Wang, Jiale Yuan, Jun Song, Bo Zheng, Linfeng Zhang, Siteng Huang, Honggang Chen
Abstract: Multimodal large language models (MLLMs) have attracted considerable attention due to their exceptional performance in visual content understanding and reasoning. However, their inference efficiency has been a notable concern, as the increasing length of multimodal contexts leads to quadratic complexity. Token compression techniques, which reduce the number of visual tokens, have demonstrated their effectiveness in reducing computational costs. Yet, these approaches have struggled to keep pace with the rapid advancements in MLLMs, especially the AnyRes strategy in the context of high-resolution image understanding. In this paper, we propose a novel token compression method, GlobalCom$^2$, tailored for high-resolution MLLMs that receive both the thumbnail and multiple crops. GlobalCom$^2$ treats the tokens derived from the thumbnail as the ``commander'' of the entire token compression process, directing the allocation of retention ratios and the specific compression for each crop. In this way, redundant tokens are eliminated while important local details are adaptively preserved to the highest extent feasible. Empirical results across 10 benchmarks reveal that GlobalCom$^2$ achieves an optimal balance between performance and efficiency, and consistently outperforms state-of-the-art token compression methods with LLaVA-NeXT-7B/13B models. Our code is released at \url{https://github.com/xuyang-liu16/GlobalCom2}.
Authors: Shaurya Singh Rathore, Aravind Shenoy, Krish Didwania, Aditya Kasliwal, Ujjwal Verma
Abstract: Recent advancements in image translation for enhancing mixed-exposure images have demonstrated the transformative potential of deep learning algorithms. However, addressing extreme exposure variations in images remains a significant challenge due to the inherent complexity and contrast inconsistencies across regions. Current methods often struggle to adapt effectively to these variations, resulting in suboptimal performance. In this work, we propose HipyrNet, a novel approach that integrates a HyperNetwork within a Laplacian Pyramid-based framework to tackle the challenges of mixed-exposure image enhancement. The inclusion of a HyperNetwork allows the model to adapt to these exposure variations. HyperNetworks dynamically generates weights for another network, allowing dynamic changes during deployment. In our model, the HyperNetwork employed is used to predict optimal kernels for Feature Pyramid decomposition, which enables a tailored and adaptive decomposition process for each input image. Our enhanced translational network incorporates multiscale decomposition and reconstruction, leveraging dynamic kernel prediction to capture and manipulate features across varying scales. Extensive experiments demonstrate that HipyrNet outperforms existing methods, particularly in scenarios with extreme exposure variations, achieving superior results in both qualitative and quantitative evaluations. Our approach sets a new benchmark for mixed-exposure image enhancement, paving the way for future research in adaptive image translation.
Authors: Xueyi Ke, Satoshi Tsutsui, Yayun Zhang, Bihan Wen
Abstract: Infants develop complex visual understanding rapidly, even preceding of the acquisition of linguistic inputs. As computer vision seeks to replicate the human vision system, understanding infant visual development may offer valuable insights. In this paper, we present an interdisciplinary study exploring this question: can a computational model that imitates the infant learning process develop broader visual concepts that extend beyond the vocabulary it has heard, similar to how infants naturally learn? To investigate this, we analyze a recently published model in Science by Vong et al.,which is trained on longitudinal, egocentric images of a single child paired with transcribed parental speech. We introduce a training-free framework that can discover visual concept neurons hidden in the model's internal representations. Our findings show that these neurons can classify objects outside its original vocabulary. Furthermore, we compare the visual representations in infant-like models with those in moder computer vision models, such as CLIP or ImageNet pre-trained model, highlighting key similarities and differences. Ultimately, our work bridges cognitive science and computer vision by analyzing the internal representations of a computational model trained on an infant's visual and linguistic inputs.
Authors: Rabin Dulal, Lihong Zheng, Muhammad Ashad Kabir
Abstract: Convolutional Neural Networks (CNNs) have drawn researchers' attention to identifying cattle using muzzle images. However, CNNs often fail to capture long-range dependencies within the complex patterns of the muzzle. The transformers handle these challenges. This inspired us to fuse the strengths of CNNs and transformers in muzzle-based cattle identification. Addition and concatenation have been the most commonly used techniques for feature fusion. However, addition fails to preserve discriminative information, while concatenation results in an increase in dimensionality. Both methods are simple operations and cannot discover the relationships or interactions between fusing features. This research aims to overcome the issues faced by addition and concatenation. This research introduces a novel approach called Multi-Head Attention Feature Fusion (MHAFF) for the first time in cattle identification. MHAFF captures relations between the different types of fusing features while preserving their originality. The experiments show that MHAFF outperformed addition and concatenation techniques and the existing cattle identification methods in accuracy on two publicly available cattle datasets. MHAFF demonstrates excellent performance and quickly converges to achieve optimum accuracy of 99.88% and 99.52% in two cattle datasets simultaneously.
Authors: Ludwic Leonard, Nils Thuerey, Ruediger Westermann
Abstract: We introduce a single-view reconstruction technique of volumetric fields in which multiple light scattering effects are omnipresent, such as in clouds. We model the unknown distribution of volumetric fields using an unconditional diffusion model trained on a novel benchmark dataset comprising 1,000 synthetically simulated volumetric density fields. The neural diffusion model is trained on the latent codes of a novel, diffusion-friendly, monoplanar representation. The generative model is used to incorporate a tailored parametric diffusion posterior sampling technique into different reconstruction tasks. A physically-based differentiable volume renderer is employed to provide gradients with respect to light transport in the latent space. This stands in contrast to classic NeRF approaches and makes the reconstructions better aligned with observed data. Through various experiments, we demonstrate single-view reconstruction of volumetric clouds at a previously unattainable quality.
Authors: Pei-Kang Lee, Jun-Cheng Chen, Ja-Ling Wu
Abstract: Out-of-distribution (OOD) detection has seen significant advancements with zero-shot approaches by leveraging the powerful Vision-Language Models (VLMs) such as CLIP. However, prior research works have predominantly focused on enhancing Far-OOD performance, while potentially compromising Near-OOD efficacy, as observed from our pilot study. To address this issue, we propose a novel strategy to enhance zero-shot OOD detection performances for both Far-OOD and Near-OOD scenarios by innovatively harnessing Large Language Models (LLMs) and VLMs. Our approach first exploit an LLM to generate superclasses of the ID labels and their corresponding background descriptions followed by feature extraction using CLIP. We then isolate the core semantic features for ID data by subtracting background features from the superclass features. The refined representation facilitates the selection of more appropriate negative labels for OOD data from a comprehensive candidate label set of WordNet, thereby enhancing the performance of zero-shot OOD detection in both scenarios. Furthermore, we introduce novel few-shot prompt tuning and visual prompt tuning to adapt the proposed framework to better align with the target distribution. Experimental results demonstrate that the proposed approach consistently outperforms current state-of-the-art methods across multiple benchmarks, with an improvement of up to 2.9% in AUROC and a reduction of up to 12.6% in FPR95. Additionally, our method exhibits superior robustness against covariate shift across different domains, further highlighting its effectiveness in real-world scenarios.
Authors: Sadhana Ravikumar, Asma A. Khan, Matthew C. Davis, Beatriz Paniagua
Abstract: External cervical resorption (ECR) is a resorptive process affecting teeth. While in some patients, active resorption ceases and gets replaced by osseous tissue, in other cases, the resorption progresses and ultimately results in tooth loss. For proper ECR assessment, cone-beam computed tomography (CBCT) is the recommended imaging modality, enabling a 3-D characterization of these lesions. While it is possible to manually identify and measure ECR resorption in CBCT scans, this process can be time intensive and highly subject to human error. Therefore, there is an urgent need to develop an automated method to identify and quantify the severity of ECR resorption using CBCT. Here, we present a method for ECR lesion segmentation that is based on automatic, binary classification of locally extracted voxel-wise texture features. We evaluate our method on 6 longitudinal CBCT datasets and show that certain texture-features can be used to accurately detect subtle CBCT signal changes due to ECR. We also present preliminary analyses clustering texture features within a lesion to stratify the defects and identify patterns indicative of calcification. These methods are important steps in developing prognostic biomarkers to predict whether ECR will continue to progress or cease, ultimately informing treatment decisions.
Authors: Zuyao You, Lingyu Kong, Lingchen Meng, Zuxuan Wu
Abstract: Foreground segmentation is a fundamental task in computer vision, encompassing various subdivision tasks. Previous research has typically designed task-specific architectures for each task, leading to a lack of unification. Moreover, they primarily focus on recognizing foreground objects without effectively distinguishing them from the background. In this paper, we emphasize the importance of the background and its relationship with the foreground. We introduce FOCUS, the Foreground ObjeCts Universal Segmentation framework that can handle multiple foreground tasks. We develop a multi-scale semantic network using the edge information of objects to enhance image features. To achieve boundary-aware segmentation, we propose a novel distillation method, integrating the contrastive learning strategy to refine the prediction mask in multi-modal feature space. We conduct extensive experiments on a total of 13 datasets across 5 tasks, and the results demonstrate that FOCUS consistently outperforms the state-of-the-art task-specific models on most metrics.
Authors: Guang Yang, Jingkun Chen, Xicheng Sheng, Shan Yang, Xiahai Zhuang, Betty Raman, Lei Li, Vicente Grau
Abstract: Late gadolinium enhancement MRI (LGE MRI) is the gold standard for the detection of myocardial scars for post myocardial infarction (MI). LGE MRI requires the injection of a contrast agent, which carries potential side effects and increases scanning time and patient discomfort. To address these issues, we propose a novel framework that combines cardiac motion observed in cine MRI with image texture information to segment the myocardium and scar tissue in the left ventricle. Cardiac motion tracking can be formulated as a full cardiac image cycle registration problem, which can be solved via deep neural networks. Experimental results prove that the proposed method can achieve scar segmentation based on non-contrasted cine images with comparable accuracy to LGE MRI. This demonstrates its potential as an alternative to contrast-enhanced techniques for scar detection.
Authors: Wen Tianci, Liu Zhiang, Lu Biao, Fang Yongchun
Abstract: 3D Gaussian Splatting (3DGS) has recently revolutionized novel view synthesis in the Simultaneous Localization and Mapping (SLAM). However, existing SLAM methods utilizing 3DGS have failed to provide high-quality novel view rendering for monocular, stereo, and RGB-D cameras simultaneously. Notably, some methods perform well for RGB-D cameras but suffer significant degradation in rendering quality for monocular cameras. In this paper, we present Scaffold-SLAM, which delivers simultaneous localization and high-quality photorealistic mapping across monocular, stereo, and RGB-D cameras. We introduce two key innovations to achieve this state-of-the-art visual quality. First, we propose Appearance-from-Motion embedding, enabling 3D Gaussians to better model image appearance variations across different camera poses. Second, we introduce a frequency regularization pyramid to guide the distribution of Gaussians, allowing the model to effectively capture finer details in the scene. Extensive experiments on monocular, stereo, and RGB-D datasets demonstrate that Scaffold-SLAM significantly outperforms state-of-the-art methods in photorealistic mapping quality, e.g., PSNR is 16.76% higher in the TUM RGB-D datasets for monocular cameras.
Authors: Shishir Muralidhara, Ren\'e Schuster, Didier Stricker
Abstract: Semantic segmentation for autonomous driving is an even more challenging task when faced with adverse driving conditions. Standard models trained on data recorded under ideal conditions show a deteriorated performance in unfavorable weather or illumination conditions. Fine-tuning on the new task or condition would lead to overwriting the previously learned information resulting in catastrophic forgetting. Adapting to the new conditions through traditional domain adaption methods improves the performance on the target domain at the expense of the source domain. Addressing these issues, we propose an architecture-based domain-incremental learning approach called Progressive Semantic Segmentation (PSS). PSS is a task-agnostic, dynamically growing collection of domain-specific segmentation models. The task of inferring the domain and subsequently selecting the appropriate module for segmentation is carried out using a collection of convolutional autoencoders. We extensively evaluate our proposed approach using several datasets at varying levels of granularity in the categorization of adverse driving conditions. Furthermore, we demonstrate the generalization of the proposed approach to similar and unseen domains.
Authors: Jiaxuan Peng, Mengshi Qi, Dong Zhao, Huadong Ma
Abstract: 3D human pose estimation (3D HPE) has emerged as a prominent research topic, particularly in the realm of RGB-based methods. However, RGB images are susceptible to limitations such as sensitivity to lighting conditions and potential user discomfort. Consequently, multi-modal sensing, which leverages non-intrusive sensors, is gaining increasing attention. Nevertheless, multi-modal 3D HPE still faces challenges, including modality imbalance and the imperative for continual learning. In this work, we introduce a novel balanced continual multi-modal learning method for 3D HPE, which harnesses the power of RGB, LiDAR, mmWave, and WiFi. Specifically, we propose a Shapley value-based contribution algorithm to quantify the contribution of each modality and identify modality imbalance. To address this imbalance, we employ a re-learning strategy. Furthermore, recognizing that raw data is prone to noise contamination, we develop a novel denoising continual learning approach. This approach incorporates a noise identification and separation module to mitigate the adverse effects of noise and collaborates with the balanced learning strategy to enhance optimization. Additionally, an adaptive EWC mechanism is employed to alleviate catastrophic forgetting. We conduct extensive experiments on the widely-adopted multi-modal dataset, MM-Fi, which demonstrate the superiority of our approach in boosting 3D pose estimation and mitigating catastrophic forgetting in complex scenarios. We will release our codes.
Authors: Wanli Ma, Oktay Karakus, Paul L. Rosin
Abstract: Cloud removal plays a crucial role in enhancing remote sensing image analysis, yet accurately reconstructing cloud-obscured regions remains a significant challenge. Recent advancements in generative models have made the generation of realistic images increasingly accessible, offering new opportunities for this task. Given the conceptual alignment between image generation and cloud removal tasks, generative models present a promising approach for addressing cloud removal in remote sensing. In this work, we propose a deep transfer learning approach built on a generative adversarial network (GAN) framework to explore the potential of the novel masked autoencoder (MAE) image reconstruction model in cloud removal. Due to the complexity of remote sensing imagery, we further propose using a patch-wise discriminator to determine whether each patch of the image is real or not. The proposed reconstructive transfer learning approach demonstrates significant improvements in cloud removal performance compared to other GAN-based methods. Additionally, whilst direct comparisons with some of the state-of-the-art cloud removal techniques are limited due to unclear details regarding their train/test data splits, the proposed model achieves competitive results based on available benchmarks.
Authors: Fabian H\"orst, Moritz Rempe, Helmut Becker, Lukas Heine, Julius Keyl, Jens Kleesiek
Abstract: Digital Pathology is a cornerstone in the diagnosis and treatment of diseases. A key task in this field is the identification and segmentation of cells in hematoxylin and eosin-stained images. Existing methods for cell segmentation often require extensive annotated datasets for training and are limited to a predefined cell classification scheme. To overcome these limitations, we propose $\text{CellViT}^{{\scriptscriptstyle ++}}$, a framework for generalized cell segmentation in digital pathology. $\text{CellViT}^{{\scriptscriptstyle ++}}$ utilizes Vision Transformers with foundation models as encoders to compute deep cell features and segmentation masks simultaneously. To adapt to unseen cell types, we rely on a computationally efficient approach. It requires minimal data for training and leads to a drastically reduced carbon footprint. We demonstrate excellent performance on seven different datasets, covering a broad spectrum of cell types, organs, and clinical settings. The framework achieves remarkable zero-shot segmentation and data-efficient cell-type classification. Furthermore, we show that $\text{CellViT}^{{\scriptscriptstyle ++}}$ can leverage immunofluorescence stainings to generate training datasets without the need for pathologist annotations. The automated dataset generation approach surpasses the performance of networks trained on manually labeled data, demonstrating its effectiveness in creating high-quality training datasets without expert annotations. To advance digital pathology, $\text{CellViT}^{{\scriptscriptstyle ++}}$ is available as an open-source framework featuring a user-friendly, web-based interface for visualization and annotation. The code is available under https://github.com/TIO-IKIM/CellViT-plus-plus.
Authors: Xinzi Cao, Xiawu Zheng, Guanhong Wang, Weijiang Yu, Yunhang Shen, Ke Li, Yutong Lu, Yonghong Tian
Abstract: Generalized Category Discovery (GCD) aims to identify a mix of known and novel categories within unlabeled data sets, providing a more realistic setting for image recognition. Essentially, GCD needs to remember existing patterns thoroughly to recognize novel categories. Recent state-of-the-art method SimGCD transfers the knowledge from known-class data to the learning of novel classes through debiased learning. However, some patterns are catastrophically forgot during adaptation and thus lead to poor performance in novel categories classification. To address this issue, we propose a novel learning approach, LegoGCD, which is seamlessly integrated into previous methods to enhance the discrimination of novel classes while maintaining performance on previously encountered known classes. Specifically, we design two types of techniques termed as Local Entropy Regularization (LER) and Dual-views Kullback Leibler divergence constraint (DKL). The LER optimizes the distribution of potential known class samples in unlabeled data, thus ensuring the preservation of knowledge related to known categories while learning novel classes. Meanwhile, DKL introduces Kullback Leibler divergence to encourage the model to produce a similar prediction distribution of two view samples from the same image. In this way, it successfully avoids mismatched prediction and generates more reliable potential known class samples simultaneously. Extensive experiments validate that the proposed LegoGCD effectively addresses the known category forgetting issue across all datasets, eg, delivering a 7.74% and 2.51% accuracy boost on known and novel classes in CUB, respectively. Our code is available at: https://github.com/Cliffia123/LegoGCD.
Authors: Nora Gourmelon, Konrad Heidler, Erik Loebel, Daniel Cheng, Julian Klink, Anda Dong, Fei Wu, Noah Maul, Moritz Koch, Marcel Dreier, Dakota Pyles, Thorsten Seehaus, Matthias Braun, Andreas Maier, Vincent Christlein
Abstract: Calving front position variation of marine-terminating glaciers is an indicator of ice mass loss and a crucial parameter in numerical glacier models. Deep Learning (DL) systems can automatically extract this position from Synthetic Aperture Radar (SAR) imagery, enabling continuous, weather- and illumination-independent, large-scale monitoring. This study presents the first comparison of DL systems on a common calving front benchmark dataset. A multi-annotator study with ten annotators is performed to contrast the best-performing DL system against human performance. The best DL model's outputs deviate 221 m on average, while the average deviation of the human annotators is 38 m. This significant difference shows that current DL systems do not yet match human performance and that further research is needed to enable fully automated monitoring of glacier calving fronts. The study of Vision Transformers, foundation models, and the inclusion and processing strategy of more information are identified as avenues for future research.
Authors: Mingzi Wang, Yuan Meng, Chen Tang, Weixiang Zhang, Yijian Qin, Yang Yao, Yingxin Li, Tongtong Feng, Xin Wang, Xun Guan, Zhi Wang, Wenwu Zhu
Abstract: The co-design of neural network architectures, quantization precisions, and hardware accelerators offers a promising approach to achieving an optimal balance between performance and efficiency, particularly for model deployment on resource-constrained edge devices. In this work, we propose the JAQ Framework, which jointly optimizes the three critical dimensions. However, effectively automating the design process across the vast search space of those three dimensions poses significant challenges, especially when pursuing extremely low-bit quantization. Specifical, the primary challenges include: (1) Memory overhead in software-side: Low-precision quantization-aware training can lead to significant memory usage due to storing large intermediate features and latent weights for back-propagation, potentially causing memory exhaustion. (2) Search time-consuming in hardware-side: The discrete nature of hardware parameters and the complex interplay between compiler optimizations and individual operators make the accelerator search time-consuming. To address these issues, JAQ mitigates the memory overhead through a channel-wise sparse quantization (CSQ) scheme, selectively applying quantization to the most sensitive components of the model during optimization. Additionally, JAQ designs BatchTile, which employs a hardware generation network to encode all possible tiling modes, thereby speeding up the search for the optimal compiler mapping strategy. Extensive experiments demonstrate the effectiveness of JAQ, achieving approximately 7% higher Top-1 accuracy on ImageNet compared to previous methods and reducing the hardware search time per iteration to 0.15 seconds.
Authors: Junha Park, Ian Ryu, Jaehui Hwang, Hyungkeun Park, Jiyoon Kim, Jong-Seok Lee
Abstract: With advances in diffusion models, image generation has shown significant performance improvements. This raises concerns about the potential abuse of image generation, such as the creation of explicit or violent images, commonly referred to as Not Safe For Work (NSFW) content. To address this, the Stable Diffusion model includes several safety checkers to censor initial text prompts and final output images generated from the model. However, recent research has shown that these safety checkers have vulnerabilities against adversarial attacks, allowing them to generate NSFW images. In this paper, we find that these adversarial attacks are not robust to small changes in text prompts or input latents. Based on this, we propose CROPS (Circular or RandOm Prompts for Safety), a model-agnostic framework that easily defends against adversarial attacks generating NSFW images without requiring additional training. Moreover, we develop an approach that utilizes one-step diffusion models for efficient NSFW detection (CROPS-1), further reducing computational resources. We demonstrate the superiority of our method in terms of performance and applicability.
Authors: Shuliang Ning, Yipeng Qin, Xiaoguang Han
Abstract: Virtual Try-On (VTON) has become a crucial tool in ecommerce, enabling the realistic simulation of garments on individuals while preserving their original appearance and pose. Early VTON methods relied on single generative networks, but challenges remain in preserving fine-grained garment details due to limitations in feature extraction and fusion. To address these issues, recent approaches have adopted a dual-network paradigm, incorporating a complementary "ReferenceNet" to enhance garment feature extraction and fusion. While effective, this dual-network approach introduces significant computational overhead, limiting its scalability for high-resolution and long-duration image/video VTON applications. In this paper, we challenge the dual-network paradigm by proposing a novel single-network VTON method that overcomes the limitations of existing techniques. Our method, namely MNVTON, introduces a Modality-specific Normalization strategy that separately processes text, image and video inputs, enabling them to share the same attention layers in a VTON network. Extensive experimental results demonstrate the effectiveness of our approach, showing that it consistently achieves higher-quality, more detailed results for both image and video VTON tasks. Our results suggest that the single-network paradigm can rival the performance of dualnetwork approaches, offering a more efficient alternative for high-quality, scalable VTON applications.
Authors: Dimitrios Gerogiannis, Foivos Paraperas Papantoniou, Rolandos Alexandros Potamias, Alexandros Lattas, Stefanos Zafeiriou
Abstract: Inspired by the effectiveness of 3D Gaussian Splatting (3DGS) in reconstructing detailed 3D scenes within multi-view setups and the emergence of large 2D human foundation models, we introduce Arc2Avatar, the first SDS-based method utilizing a human face foundation model as guidance with just a single image as input. To achieve that, we extend such a model for diverse-view human head generation by fine-tuning on synthetic data and modifying its conditioning. Our avatars maintain a dense correspondence with a human face mesh template, allowing blendshape-based expression generation. This is achieved through a modified 3DGS approach, connectivity regularizers, and a strategic initialization tailored for our task. Additionally, we propose an optional efficient SDS-based correction step to refine the blendshape expressions, enhancing realism and diversity. Experiments demonstrate that Arc2Avatar achieves state-of-the-art realism and identity preservation, effectively addressing color issues by allowing the use of very low guidance, enabled by our strong identity prior and initialization strategy, without compromising detail.
Authors: Athulya Sundaresan Geetha
Abstract: Safe knife practices in the kitchen significantly reduce the risk of cuts, injuries, and serious accidents during food preparation. Using YOLOv7, an advanced object detection model, this study focuses on identifying safety risks during knife handling, particularly improper finger placement and blade contact with hand. The model's performance was evaluated using metrics such as precision, recall, mAP50, and mAP50-95. The results demonstrate that YOLOv7 achieved its best performance at epoch 31, with a mAP50-95 score of 0.7879, precision of 0.9063, and recall of 0.7503. These findings highlight YOLOv7's potential to accurately detect knife-related hazards, promoting the development of improved kitchen safety.
Authors: Maximilian Alber, Stephan Tietz, Jonas Dippel, Timo Milbich, Timoth\'ee Lesort, Panos Korfiatis, Moritz Kr\"ugener, Beatriz Perez Cancer, Neelay Shah, Alexander M\"ollers, Philipp Seegerer, Alexandra Carpen-Amarie, Kai Standvoss, Gabriel Dernbach, Edwin de Jong, Simon Schallenberg, Andreas Kunft, Helmut Hoffer von Ankershoffen, Gavin Schaeferle, Patrick Duffy, Matt Redlon, Philipp Jurmeister, David Horst, Lukas Ruff, Klaus-Robert M\"uller, Frederick Klauschen, Andrew Norgan
Abstract: Recent advances in digital pathology have demonstrated the effectiveness of foundation models across diverse applications. In this report, we present a novel vision foundation model based on the RudolfV approach. Our model was trained on a dataset comprising 1.2 million histopathology whole slide images, collected from two medical institutions: Mayo Clinic and Charit\'e - Universt\"atsmedizin Berlin. Comprehensive evaluations show that our model achieves state-of-the-art performance across twenty-one public benchmark datasets, even though it is neither the largest model by parameter count nor by training dataset size.
Authors: Md. Arafat Alam Khandaker, Ziyan Shirin Raha, Salehin Bin Iqbal, M. F. Mridha, Jungpil Shin
Abstract: Brain cancer represents a major challenge in medical diagnostics, requisite precise and timely detection for effective treatment. Diagnosis initially relies on the proficiency of radiologists, which can cause difficulties and threats when the expertise is sparse. Despite the use of imaging resources, brain cancer remains often difficult, time-consuming, and vulnerable to intraclass variability. This study conveys the Bangladesh Brain Cancer MRI Dataset, containing 6,056 MRI images organized into three categories: Brain Tumor, Brain Glioma, and Brain Menin. The dataset was collected from several hospitals in Bangladesh, providing a diverse and realistic sample for research. We implemented advanced deep learning models, and DenseNet169 achieved exceptional results, with accuracy, precision, recall, and F1-Score all reaching 0.9983. In addition, Explainable AI (XAI) methods including GradCAM, GradCAM++, ScoreCAM, and LayerCAM were employed to provide visual representations of the decision-making processes of the models. In the context of brain cancer, these techniques highlight DenseNet169's potential to enhance diagnostic accuracy while simultaneously offering transparency, facilitating early diagnosis and better patient outcomes.
Authors: Xuyi Meng, Chen Wang, Jiahui Lei, Kostas Daniilidis, Jiatao Gu, Lingjie Liu
Abstract: Recent advances in 2D image generation have achieved remarkable quality,largely driven by the capacity of diffusion models and the availability of large-scale datasets. However, direct 3D generation is still constrained by the scarcity and lower fidelity of 3D datasets. In this paper, we introduce Zero-1-to-G, a novel approach that addresses this problem by enabling direct single-view generation on Gaussian splats using pretrained 2D diffusion models. Our key insight is that Gaussian splats, a 3D representation, can be decomposed into multi-view images encoding different attributes. This reframes the challenging task of direct 3D generation within a 2D diffusion framework, allowing us to leverage the rich priors of pretrained 2D diffusion models. To incorporate 3D awareness, we introduce cross-view and cross-attribute attention layers, which capture complex correlations and enforce 3D consistency across generated splats. This makes Zero-1-to-G the first direct image-to-3D generative model to effectively utilize pretrained 2D diffusion priors, enabling efficient training and improved generalization to unseen objects. Extensive experiments on both synthetic and in-the-wild datasets demonstrate superior performance in 3D object generation, offering a new approach to high-quality 3D generation.
Authors: Maxime Dieudonn\'e (Institut de Neurosciences de la Timone, UMR 7289, CNRS, Aix-Marseille Universit\'e, 13005, Marseille, France), Guillaume Auzias (Institut de Neurosciences de la Timone, UMR 7289, CNRS, Aix-Marseille Universit\'e, 13005, Marseille, France), Julien Lef\`evre (Institut de Neurosciences de la Timone, UMR 7289, CNRS, Aix-Marseille Universit\'e, 13005, Marseille, France)
Abstract: The shape of human brain is complex and highly variable, with interactions between brain size, cortical folding, and age well-documented in the literature. However, few studies have explored how global brain size influences geometric features of the cortical surface derived from anatomical MRI. In this work, we focus on sulcal depth, an imaging phenotype that has gained significant attention in both basic research and clinical applications. We make key contributions to the field by: 1) providing the first quantitative analysis of how brain size affects sulcal depth measurements; 2) introducing a novel, scale-invariant method for sulcal depth estimation based on an original formalization of the problem; 3) presenting a validation framework and sharing our code and benchmark data with the community; and 4) demonstrating the biological relevance of our new sulcal depth measure using a large sample of 1,987 subjects spanning the developmental period from 26 weeks post-conception to adulthood.
Authors: Aniruddha Mahapatra, Long Mai, Yitian Zhang, David Bourgin, Feng Liu
Abstract: Video tokenizers are essential for latent video diffusion models, converting raw video data into spatiotemporally compressed latent spaces for efficient training. However, extending state-of-the-art video tokenizers to achieve a temporal compression ratio beyond 4x without increasing channel capacity poses significant challenges. In this work, we propose an alternative approach to enhance temporal compression. We find that the reconstruction quality of temporally subsampled videos from a low-compression encoder surpasses that of high-compression encoders applied to original videos. This indicates that high-compression models can leverage representations from lower-compression models. Building on this insight, we develop a bootstrapped high-temporal-compression model that progressively trains high-compression blocks atop well-trained lower-compression models. Our method includes a cross-level feature-mixing module to retain information from the pretrained low-compression model and guide higher-compression blocks to capture the remaining details from the full video sequence. Evaluation of video benchmarks shows that our method significantly improves reconstruction quality while increasing temporal compression compared to direct extensions of existing video tokenizers. Furthermore, the resulting compact latent space effectively trains a video diffusion model for high-quality video generation with a reduced token budget.
Authors: Yunzhuo Hao, Jiawei Gu, Huichen Will Wang, Linjie Li, Zhengyuan Yang, Lijuan Wang, Yu Cheng
Abstract: The ability to organically reason over and with both text and images is a pillar of human intelligence, yet the ability of Multimodal Large Language Models (MLLMs) to perform such multimodal reasoning remains under-explored. Existing benchmarks often emphasize text-dominant reasoning or rely on shallow visual cues, failing to adequately assess integrated visual and textual reasoning. We introduce EMMA (Enhanced MultiModal reAsoning), a benchmark targeting organic multimodal reasoning across mathematics, physics, chemistry, and coding. EMMA tasks demand advanced cross-modal reasoning that cannot be addressed by reasoning independently in each modality, offering an enhanced test suite for MLLMs' reasoning capabilities. Our evaluation of state-of-the-art MLLMs on EMMA reveals significant limitations in handling complex multimodal and multi-step reasoning tasks, even with advanced techniques like Chain-of-Thought prompting and test-time compute scaling underperforming. These findings underscore the need for improved multimodal architectures and training paradigms to close the gap between human and model reasoning in multimodality.
Authors: Runjie Yan, Yinbo Chen, Xiaolong Wang
Abstract: Score Distillation Sampling (SDS) has made significant strides in distilling image-generative models for 3D generation. However, its maximum-likelihood-seeking behavior often leads to degraded visual quality and diversity, limiting its effectiveness in 3D applications. In this work, we propose Consistent Flow Distillation (CFD), which addresses these limitations. We begin by leveraging the gradient of the diffusion ODE or SDE sampling process to guide the 3D generation. From the gradient-based sampling perspective, we find that the consistency of 2D image flows across different viewpoints is important for high-quality 3D generation. To achieve this, we introduce multi-view consistent Gaussian noise on the 3D object, which can be rendered from various viewpoints to compute the flow gradient. Our experiments demonstrate that CFD, through consistent flows, significantly outperforms previous methods in text-to-3D generation.
Authors: Yifan Yu, Shaohui Liu, R\'emi Pautrat, Marc Pollefeys, Viktor Larsson
Abstract: Monocular depth estimation (MDE) models have undergone significant advancements over recent years. Many MDE models aim to predict affine-invariant relative depth from monocular images, while recent developments in large-scale training and vision foundation models enable reasonable estimation of metric (absolute) depth. However, effectively leveraging these predictions for geometric vision tasks, in particular relative pose estimation, remains relatively under explored. While depths provide rich constraints for cross-view image alignment, the intrinsic noise and ambiguity from the monocular depth priors present practical challenges to improving upon classic keypoint-based solutions. In this paper, we develop three solvers for relative pose estimation that explicitly account for independent affine (scale and shift) ambiguities, covering both calibrated and uncalibrated conditions. We further propose a hybrid estimation pipeline that combines our proposed solvers with classic point-based solvers and epipolar constraints. We find that the affine correction modeling is beneficial to not only the relative depth priors but also, surprisingly, the ``metric" ones. Results across multiple datasets demonstrate large improvements of our approach over classic keypoint-based baselines and PnP-based solutions, under both calibrated and uncalibrated setups. We also show that our method improves consistently with different feature matchers and MDE models, and can further benefit from very recent advances on both modules. Code is available at https://github.com/MarkYu98/madpose.
Authors: Md. Arafat Alam Khandaker, Ziyan Shirin Raha, Shifat Islam, Tashreef Muhammad
Abstract: Pumpkin leaf diseases are significant threats to agricultural productivity, requiring a timely and precise diagnosis for effective management. Traditional identification methods are laborious and susceptible to human error, emphasizing the necessity for automated solutions. This study employs on the "Pumpkin Leaf Disease Dataset", that comprises of 2000 high-resolution images separated into five categories. Downy mildew, powdery mildew, mosaic disease, bacterial leaf spot, and healthy leaves. The dataset was rigorously assembled from several agricultural fields to ensure a strong representation for model training. We explored many proficient deep learning architectures, including DenseNet201, DenseNet121, DenseNet169, Xception, ResNet50, ResNet101 and InceptionResNetV2, and observed that ResNet50 performed most effectively, with an accuracy of 90.5% and comparable precision, recall, and F1-Score. We used Explainable AI (XAI) approaches like Grad-CAM, Grad-CAM++, Score-CAM, and Layer-CAM to provide meaningful representations of model decision-making processes, which improved understanding and trust in automated disease diagnostics. These findings demonstrate ResNet50's potential to revolutionize pumpkin leaf disease detection, allowing for earlier and more accurate treatments.
Authors: David McAllister, Matthew Tancik, Jiaming Song, Angjoo Kanazawa
Abstract: Large-scale AI model training divides work across thousands of GPUs, then synchronizes gradients across them at each step. This incurs a significant network burden that only centralized, monolithic clusters can support, driving up infrastructure costs and straining power systems. We propose Decentralized Diffusion Models, a scalable framework for distributing diffusion model training across independent clusters or datacenters by eliminating the dependence on a centralized, high-bandwidth networking fabric. Our method trains a set of expert diffusion models over partitions of the dataset, each in full isolation from one another. At inference time, the experts ensemble through a lightweight router. We show that the ensemble collectively optimizes the same objective as a single model trained over the whole dataset. This means we can divide the training burden among a number of "compute islands," lowering infrastructure costs and improving resilience to localized GPU failures. Decentralized diffusion models empower researchers to take advantage of smaller, more cost-effective and more readily available compute like on-demand GPU nodes rather than central integrated systems. We conduct extensive experiments on ImageNet and LAION Aesthetics, showing that decentralized diffusion models FLOP-for-FLOP outperform standard diffusion models. We finally scale our approach to 24 billion parameters, demonstrating that high-quality diffusion models can now be trained with just eight individual GPU nodes in less than a week.
Authors: Xingyu Fu, Minqian Liu, Zhengyuan Yang, John Corring, Yijuan Lu, Jianwei Yang, Dan Roth, Dinei Florencio, Cha Zhang
Abstract: Structured image understanding, such as interpreting tables and charts, requires strategically refocusing across various structures and texts within an image, forming a reasoning sequence to arrive at the final answer. However, current multimodal large language models (LLMs) lack this multihop selective attention capability. In this work, we introduce ReFocus, a simple yet effective framework that equips multimodal LLMs with the ability to generate "visual thoughts" by performing visual editing on the input image through code, shifting and refining their visual focuses. Specifically, ReFocus enables multimodal LLMs to generate Python codes to call tools and modify the input image, sequentially drawing boxes, highlighting sections, and masking out areas, thereby enhancing the visual reasoning process. We experiment upon a wide range of structured image understanding tasks involving tables and charts. ReFocus largely improves performance on all tasks over GPT-4o without visual editing, yielding an average gain of 11.0% on table tasks and 6.8% on chart tasks. We present an in-depth analysis of the effects of different visual edits, and reasons why ReFocus can improve the performance without introducing additional information. Further, we collect a 14k training set using ReFocus, and prove that such visual chain-of-thought with intermediate information offers a better supervision than standard VQA data, reaching a 8.0% average gain over the same model trained with QA pairs and 2.6% over CoT.
Authors: Jathushan Rajasegaran, Ilija Radosavovic, Rahul Ravishankar, Yossi Gandelsman, Christoph Feichtenhofer, Jitendra Malik
Abstract: We empirically study autoregressive pre-training from videos. To perform our study, we construct a series of autoregressive video models, called Toto. We treat videos as sequences of visual tokens and train transformer models to autoregressively predict future tokens. Our models are pre-trained on a diverse dataset of videos and images comprising over 1 trillion visual tokens. We explore different architectural, training, and inference design choices. We evaluate the learned visual representations on a range of downstream tasks including image recognition, video classification, object tracking, and robotics. Our results demonstrate that, despite minimal inductive biases, autoregressive pre-training leads to competitive performance across all benchmarks. Finally, we find that scaling our video models results in similar scaling curves to those seen in language models, albeit with a different rate. More details at https://brjathu.github.io/toto/
Authors: J. Yu, H. Yi, Y. Wang, J. D. Opfermann, W. G. Gensheimer, A. Krieger, J. U. Kang
Abstract: Deep Anterior Lamellar Keratoplasty (DALK) is a partial-thickness corneal transplant procedure used to treat corneal stromal diseases. A crucial step in this procedure is the precise separation of the deep stroma from Descemet's membrane (DM) using the Big Bubble technique. To simplify the tasks of needle insertion and pneumo-dissection in this technique, we previously developed an Optical Coherence Tomography (OCT)-guided, eye-mountable robot that uses real-time tracking of corneal layers from M-mode OCT signals for control. However, signal noise and instability during manipulation of the OCT fiber sensor-integrated needle have hindered the performance of conventional deep-learning segmentation methods, resulting in rough and inaccurate detection of corneal layers. To address these challenges, we have developed a topology-based deep-learning segmentation method that integrates a topological loss function with a modified network architecture. This approach effectively reduces the effects of noise and improves segmentation speed, precision, and stability. Validation using in vivo, ex vivo, and hybrid rabbit eye datasets demonstrates that our method outperforms traditional loss-based techniques, providing fast, accurate, and robust segmentation of the epithelium and DM to guide surgery.
Authors: Gianfranco Cortes, Baba C. Vemuri
Abstract: Nonrigid registration is vital to medical image analysis but remains challenging for diffusion MRI (dMRI) due to its high-dimensional, orientation-dependent nature. While classical methods are accurate, they are computationally demanding, and deep neural networks, though efficient, have been underexplored for nonrigid dMRI registration compared to structural imaging. We present a novel, deep learning framework for model-free, nonrigid registration of raw diffusion MRI data that does not require explicit reorientation. Unlike previous methods relying on derived representations such as diffusion tensors or fiber orientation distribution functions, in our approach, we formulate the registration as an equivariant diffeomorphism of position-and-orientation space. Central to our method is an $\mathsf{SE}(3)$-equivariant UNet that generates velocity fields while preserving the geometric properties of a raw dMRI's domain. We introduce a new loss function based on the maximum mean discrepancy in Fourier space, implicitly matching ensemble average propagators across images. Experimental results on Human Connectome Project dMRI data demonstrate competitive performance compared to state-of-the-art approaches, with the added advantage of bypassing the overhead for estimating derived representations. This work establishes a foundation for data-driven, geometry-aware dMRI registration directly in the acquisition space.
Authors: Haoran Zhu, Zhenyuan Dong, Kristi Topollai, Anna Choromanska
Abstract: As opposed to human drivers, current autonomous driving systems still require vast amounts of labeled data to train. Recently, world models have been proposed to simultaneously enhance autonomous driving capabilities by improving the way these systems understand complex real-world environments and reduce their data demands via self-supervised pre-training. In this paper, we present AD-L-JEPA (aka Autonomous Driving with LiDAR data via a Joint Embedding Predictive Architecture), a novel self-supervised pre-training framework for autonomous driving with LiDAR data that, as opposed to existing methods, is neither generative nor contrastive. Our method learns spatial world models with a joint embedding predictive architecture. Instead of explicitly generating masked unknown regions, our self-supervised world models predict Bird's Eye View (BEV) embeddings to represent the diverse nature of autonomous driving scenes. Our approach furthermore eliminates the need to manually create positive and negative pairs, as is the case in contrastive learning. AD-L-JEPA leads to simpler implementation and enhanced learned representations. We qualitatively and quantitatively demonstrate high-quality of embeddings learned with AD-L-JEPA. We furthermore evaluate the accuracy and label efficiency of AD-L-JEPA on popular downstream tasks such as LiDAR 3D object detection and associated transfer learning. Our experimental evaluation demonstrates that AD-L-JEPA is a plausible approach for self-supervised pre-training in autonomous driving applications and is the best available approach outperforming SOTA, including most recently proposed Occupancy-MAE [1] and ALSO [2]. The source code of AD-L-JEPA is available at https://github.com/HaoranZhuExplorer/AD-L-JEPA-Release.
URLs: https://github.com/HaoranZhuExplorer/AD-L-JEPA-Release.
Authors: Toshit Jain, Upkar Singh, Varun Singh, Vijay Kumar Boda, Ingrid Hotz, Sathish S. Vadhiyar, P. N. Vinayachandran, Vijay Natarajan
Abstract: Oceanographers rely on visual analysis to interpret model simulations, identify events and phenomena, and track dynamic ocean processes. The ever increasing resolution and complexity of ocean data due to its dynamic nature and multivariate relationships demands a scalable and adaptable visualization tool for interactive exploration. We introduce pyParaOcean, a scalable and interactive visualization system designed specifically for ocean data analysis. pyParaOcean offers specialized modules for common oceanographic analysis tasks, including eddy identification and salinity movement tracking. These modules seamlessly integrate with ParaView as filters, ensuring a user-friendly and easy-to-use system while leveraging the parallelization capabilities of ParaView and a plethora of inbuilt general-purpose visualization functionalities. The creation of an auxiliary dataset stored as a Cinema database helps address I/O and network bandwidth bottlenecks while supporting the generation of quick overview visualizations. We present a case study on the Bay of Bengal (BoB) to demonstrate the utility of the system and scaling studies to evaluate the efficiency of the system.
Authors: Chongzhi Zhang, Xizhou Zhu, Aixin Sun
Abstract: Video moment search, the process of finding relevant moments in a video corpus to match a user's query, is crucial for various applications. Existing solutions, however, often assume a single perfect matching moment, struggle with inefficient inference, and have limitations with hour-long videos. This paper introduces a flexible and scalable framework for retrieving a ranked list of moments from collection of videos in any length to match a text query, a task termed Ranked Video Moment Retrieval (RVMR). Our framework, called Segment-Proposal-Ranking (SPR), simplifies the search process into three independent stages: segment retrieval, proposal generation, and moment refinement with re-ranking. Specifically, videos are divided into equal-length segments with precomputed embeddings indexed offline, allowing efficient retrieval regardless of video length. For scalable online retrieval, both segments and queries are projected into a shared feature space to enable approximate nearest neighbor (ANN) search. Retrieved segments are then merged into coarse-grained moment proposals. Then a refinement and re-ranking module is designed to reorder and adjust timestamps of the coarse-grained proposals. Evaluations on the TVR-Ranking dataset demonstrate that our framework achieves state-of-the-art performance with significant reductions in computational cost and processing time. The flexible design also allows for independent improvements to each stage, making SPR highly adaptable for large-scale applications.
Authors: Andrei Iantsen
Abstract: Tumor volume segmentation on MRI is a challenging and time-consuming process that is performed manually in typical clinical settings. This work presents an approach to automated delineation of head and neck tumors on MRI scans, developed in the context of the MICCAI Head and Neck Tumor Segmentation for MR-Guided Applications (HNTS-MRG) 2024 Challenge. Rather than designing a new, task-specific convolutional neural network, the focus of this research was to propose improvements to the configuration commonly used in medical segmentation tasks, relying solely on the traditional U-Net architecture. The empirical results presented in this article suggest the superiority of patch-wise normalization used for both training and sliding window inference. They also indicate that the performance of segmentation models can be enhanced by applying a scheduled data augmentation policy during training. Finally, it is shown that a small improvement in quality can be achieved by using Gaussian weighting to combine predictions for individual patches during sliding window inference. The model with the best configuration obtained an aggregated Dice Similarity Coefficient (DSCagg) of 0.749 in Task 1 and 0.710 in Task 2 on five cross-validation folds. The ensemble of five models (one best model per validation fold) showed consistent results on a private test set of 50 patients with an DSCagg of 0.752 in Task 1 and 0.718 in Task 2 (team name: andrei.iantsen). The source code and model weights are freely available at www.github.com/iantsen/hntsmrg.
Authors: Gregor Geigle, Florian Schneider, Carolin Holtermann, Chris Biemann, Radu Timofte, Anne Lauscher, Goran Glava\v{s}
Abstract: Most Large Vision-Language Models (LVLMs) to date are trained predominantly on English data, which makes them struggle to understand non-English input and fail to generate output in the desired target language. Existing efforts mitigate these issues by adding multilingual training data, but do so in a largely ad-hoc manner, lacking insight into how different training mixes tip the scale for different groups of languages. In this work, we present a comprehensive investigation into the training strategies for massively multilingual LVLMs. First, we conduct a series of multi-stage experiments spanning 13 downstream vision-language tasks and 43 languages, systematically examining: (1) the number of training languages that can be included without degrading English performance and (2) optimal language distributions of pre-training as well as (3) instruction-tuning data. Further, we (4) investigate how to improve multilingual text-in-image understanding, and introduce a new benchmark for the task. Surprisingly, our analysis reveals that one can (i) include as many as 100 training languages simultaneously (ii) with as little as 25-50\% of non-English data, to greatly improve multilingual performance while retaining strong English performance. We further find that (iii) including non-English OCR data in pre-training and instruction-tuning is paramount for improving multilingual text-in-image understanding. Finally, we put all our findings together and train Centurio, a 100-language LVLM, offering state-of-the-art performance in an evaluation covering 14 tasks and 56 languages.
Authors: Wenhao Liao, Sineng Yan, Youqian Zhang, Xinwei Zhai, Yuanyuan Wang, Eugene Yujun Fu
Abstract: Autonomous vehicles rely on camera-based perception systems to comprehend their driving environment and make crucial decisions, thereby ensuring vehicles to steer safely. However, a significant threat known as Electromagnetic Signal Injection Attacks (ESIA) can distort the images captured by these cameras, leading to incorrect AI decisions and potentially compromising the safety of autonomous vehicles. Despite the serious implications of ESIA, there is limited understanding of its impacts on the robustness of AI models across various and complex driving scenarios. To address this gap, our research analyzes the performance of different models under ESIA, revealing their vulnerabilities to the attacks. Moreover, due to the challenges in obtaining real-world attack data, we develop a novel ESIA simulation method and generate a simulated attack dataset for different driving scenarios. Our research provides a comprehensive simulation and evaluation framework, aiming to enhance the development of more robust AI models and secure intelligent systems, ultimately contributing to the advancement of safer and more reliable technology across various fields.
Authors: Talha Sultan, Alex Bocchieri, Chaoying Gu, Xiaochun Liu, Pavel Polynkin, Andreas Velten
Abstract: Non-line-of-Sight (NLOS) imaging systems collect light at a diffuse relay surface and input this measurement into computational algorithms that output a 3D volumetric reconstruction. These algorithms utilize the Fast Fourier Transform (FFT) to accelerate the reconstruction process but require both input and output to be sampled spatially with uniform grids. However, the geometry of NLOS imaging inherently results in non-uniform sampling on the relay surface when using multi-pixel detector arrays, even though such arrays significantly reduce acquisition times. Furthermore, using these arrays increases the data rate required for sensor readout, posing challenges for real-world deployment. In this work, we utilize the phasor field framework to demonstrate that existing NLOS imaging setups typically oversample the relay surface spatially, explaining why the measurement can be compressed without significantly sacrificing reconstruction quality. This enables us to utilize the Non-Uniform Fast Fourier Transform (NUFFT) to reconstruct from sparse measurements acquired from irregularly sampled relay surfaces of arbitrary shapes. Furthermore, we utilize the NUFFT to reconstruct at arbitrary locations in the hidden volume, ensuring flexible sampling schemes for both the input and output. Finally, we utilize the Scaled Fast Fourier Transform (SFFT) to reconstruct larger volumes without increasing the number of samples stored in memory. All algorithms introduced in this paper preserve the computational complexity of FFT-based methods, ensuring scalability for practical NLOS imaging applications.
Authors: Darius Petermann, Mahdi M. Kalayeh
Abstract: Training audio-to-image generative models requires an abundance of diverse audio-visual pairs that are semantically aligned. Such data is almost always curated from in-the-wild videos, given the cross-modal semantic correspondence that is inherent to them. In this work, we hypothesize that insisting on the absolute need for ground truth audio-visual correspondence, is not only unnecessary, but also leads to severe restrictions in scale, quality, and diversity of the data, ultimately impairing its use in the modern generative models. That is, we propose a scalable image sonification framework where instances from a variety of high-quality yet disjoint uni-modal origins can be artificially paired through a retrieval process that is empowered by reasoning capabilities of modern vision-language models. To demonstrate the efficacy of this approach, we use our sonified images to train an audio-to-image generative model that performs competitively against state-of-the-art. Finally, through a series of ablation studies, we exhibit several intriguing auditory capabilities like semantic mixing and interpolation, loudness calibration and acoustic space modeling through reverberation that our model has implicitly developed to guide the image generation process.
Authors: Sameer Agarwal, Erin Connelly, Annalisa Crannell, Timothy Duff, Rekha R. Thomas
Abstract: When is it possible to project two sets of labeled points lying in a pair of projective planes to the same image on a projective line? We give a complete answer to this question and describe the loci of the projection centers that enable a common image. In particular, we find that there exists a solution to this problem if and only if these two sets are themselves images of a common pointset in projective space.
Authors: Yiwen Huang, Aaron Gokaslan, Volodymyr Kuleshov, James Tompkin
Abstract: There is a widely-spread claim that GANs are difficult to train, and GAN architectures in the literature are littered with empirical tricks. We provide evidence against this claim and build a modern GAN baseline in a more principled manner. First, we derive a well-behaved regularized relativistic GAN loss that addresses issues of mode dropping and non-convergence that were previously tackled via a bag of ad-hoc tricks. We analyze our loss mathematically and prove that it admits local convergence guarantees, unlike most existing relativistic losses. Second, our new loss allows us to discard all ad-hoc tricks and replace outdated backbones used in common GANs with modern architectures. Using StyleGAN2 as an example, we present a roadmap of simplification and modernization that results in a new minimalist baseline -- R3GAN. Despite being simple, our approach surpasses StyleGAN2 on FFHQ, ImageNet, CIFAR, and Stacked MNIST datasets, and compares favorably against state-of-the-art GANs and diffusion models.
Authors: David Weber, Florian Merkle, Pascal Sch\"ottle, Stephan Schl\"ogl
Abstract: Modern, state-of-the-art Convolutional Neural Networks (CNNs) in computer vision have millions of parameters. Thus, explaining the complex decisions of such networks to humans is challenging. A technical approach to reduce CNN complexity is network pruning, where less important parameters are deleted. The work presented in this paper investigates whether this technical complexity reduction also helps with perceived explainability. To do so, we conducted a pre-study and two human-grounded experiments, assessing the effects of different pruning ratios on CNN explainability. Overall, we evaluated four different compression rates (i.e., CPR 2, 4, 8, and 32) with 37 500 tasks on Mechanical Turk. Results indicate that lower compression rates have a positive influence on explainability, while higher compression rates show negative effects. Furthermore, we were able to identify sweet spots that increase both the perceived explainability and the model's performance.
Authors: Sheng Zhang, Yanbo Xu, Naoto Usuyama, Hanwen Xu, Jaspreet Bagga, Robert Tinn, Sam Preston, Rajesh Rao, Mu Wei, Naveen Valluri, Cliff Wong, Andrea Tupini, Yu Wang, Matt Mazzola, Swadheen Shukla, Lars Liden, Jianfeng Gao, Angela Crabtree, Brian Piening, Carlo Bifulco, Matthew P. Lungren, Tristan Naumann, Sheng Wang, Hoifung Poon
Abstract: Biomedical data is inherently multimodal, comprising physical measurements and natural language narratives. A generalist biomedical AI model needs to simultaneously process different modalities of data, including text and images. Therefore, training an effective generalist biomedical model requires high-quality multimodal data, such as parallel image-text pairs. Here, we present PMC-15M, a novel dataset that is two orders of magnitude larger than existing biomedical multimodal datasets such as MIMIC-CXR, and spans a diverse range of biomedical image types. PMC-15M contains 15 million biomedical image-text pairs collected from 4.4 million scientific articles. Based on PMC-15M, we have pretrained BiomedCLIP, a multimodal foundation model, with domain-specific adaptations tailored to biomedical vision-language processing. We conducted extensive experiments and ablation studies on standard biomedical imaging tasks from retrieval to classification to visual question-answering (VQA). BiomedCLIP achieved new state-of-the-art results in a wide range of standard datasets, substantially outperforming prior approaches. Intriguingly, by large-scale pretraining on diverse biomedical image types, BiomedCLIP even outperforms state-of-the-art radiology-specific models such as BioViL in radiology-specific tasks such as RSNA pneumonia detection. In summary, BiomedCLIP is a fully open-access foundation model that achieves state-of-the-art performance on various biomedical tasks, paving the way for transformative multimodal biomedical discovery and applications. We release our models at https://aka.ms/biomedclip to facilitate future research in multimodal biomedical AI.
Authors: Tao Sun, Yan Hao, Shengyu Huang, Silvio Savarese, Konrad Schindler, Marc Pollefeys, Iro Armeni
Abstract: Building 3D geometric maps of man-made spaces is a well-established and active field that is fundamental to computer vision and robotics. However, considering the evolving nature of built environments, it is essential to question the capabilities of current mapping efforts in handling temporal changes. In addition, spatiotemporal mapping holds significant potential for achieving sustainability and circularity goals. Existing mapping approaches focus on small changes, such as object relocation or self-driving car operation; in all cases where the main structure of the scene remains fixed. Consequently, these approaches fail to address more radical changes in the structure of the built environment, such as geometry and topology. To this end, we introduce the Nothing Stands Still (NSS) benchmark, which focuses on the spatiotemporal registration of 3D scenes undergoing large spatial and temporal change, ultimately creating one coherent spatiotemporal map. Specifically, the benchmark involves registering two or more partial 3D point clouds (fragments) from the same scene but captured from different spatiotemporal views. In addition to the standard pairwise registration, we assess the multi-way registration of multiple fragments that belong to any temporal stage. As part of NSS, we introduce a dataset of 3D point clouds recurrently captured in large-scale building indoor environments that are under construction or renovation. The NSS benchmark presents three scenarios of increasing difficulty, to quantify the generalization ability of point cloud registration methods over space (within one building and across buildings) and time. We conduct extensive evaluations of state-of-the-art methods on NSS. The results demonstrate the necessity for novel methods specifically designed to handle large spatiotemporal changes. The homepage of our benchmark is at http://nothing-stands-still.com.
Authors: Jiaming Han, Kaixiong Gong, Yiyuan Zhang, Jiaqi Wang, Kaipeng Zhang, Dahua Lin, Yu Qiao, Peng Gao, Xiangyu Yue
Abstract: Multimodal large language models (MLLMs) have gained significant attention due to their strong multimodal understanding capability. However, existing works rely heavily on modality-specific encoders, which usually differ in architecture and are limited to common modalities. In this paper, we present OneLLM, an MLLM that aligns eight modalities to language using a unified framework. We achieve this through a unified multimodal encoder and a progressive multimodal alignment pipeline. In detail, we first train an image projection module to connect a vision encoder with LLM. Then, we build a universal projection module (UPM) by mixing multiple image projection modules and dynamic routing. Finally, we progressively align more modalities to LLM with the UPM. To fully leverage the potential of OneLLM in following instructions, we also curated a comprehensive multimodal instruction dataset, including 2M items from image, audio, video, point cloud, depth/normal map, IMU and fMRI brain activity. OneLLM is evaluated on 25 diverse benchmarks, encompassing tasks such as multimodal captioning, question answering and reasoning, where it delivers excellent performance. Code, data, model and online demo are available at https://github.com/csuhan/OneLLM
Authors: Donglin Di, Jiahui Yang, Chaofan Luo, Zhou Xue, Wei Chen, Xun Yang, Yue Gao
Abstract: Text-to-3D generation represents an exciting field that has seen rapid advancements, facilitating the transformation of textual descriptions into detailed 3D models. However, current progress often neglects the intricate high-order correlation of geometry and texture within 3D objects, leading to challenges such as over-smoothness, over-saturation and the Janus problem. In this work, we propose a method named ``3D Gaussian Generation via Hypergraph (Hyper-3DG)'', designed to capture the sophisticated high-order correlations present within 3D objects. Our framework is anchored by a well-established mainflow and an essential module, named ``Geometry and Texture Hypergraph Refiner (HGRefiner)''. This module not only refines the representation of 3D Gaussians but also accelerates the update process of these 3D Gaussians by conducting the Patch-3DGS Hypergraph Learning on both explicit attributes and latent visual features. Our framework allows for the production of finely generated 3D objects within a cohesive optimization, effectively circumventing degradation. Extensive experimentation has shown that our proposed method significantly enhances the quality of 3D generation while incurring no additional computational overhead for the underlying framework. (Project code: https://github.com/yjhboy/Hyper3DG)
Authors: Fan Yang, Jianfeng Zhang, Yichun Shi, Bowen Chen, Chenxu Zhang, Huichao Zhang, Xiaofeng Yang, Xiu Li, Jiashi Feng, Guosheng Lin
Abstract: Benefiting from the rapid development of 2D diffusion models, 3D content generation has witnessed significant progress. One promising solution is to finetune the pre-trained 2D diffusion models to produce multi-view images and then reconstruct them into 3D assets via feed-forward sparse-view reconstruction models. However, limited by the 3D inconsistency in the generated multi-view images and the low reconstruction resolution of the feed-forward reconstruction models, the generated 3d assets are still limited to incorrect geometries and blurry textures. To address this problem, we present a multi-view based refine method, named Magic-Boost, to further refine the generation results. In detail, we first propose a novel multi-view conditioned diffusion model which extracts 3d prior from the synthesized multi-view images to synthesize high-fidelity novel view images and then introduce a novel iterative-update strategy to adopt it to provide precise guidance to refine the coarse generated results through a fast optimization process. Conditioned on the strong 3d priors extracted from the synthesized multi-view images, Magic-Boost is capable of providing precise optimization guidance that well aligns with the coarse generated 3D assets, enriching the local detail in both geometry and texture within a short time ($\sim15$min). Extensive experiments show Magic-Boost greatly enhances the coarse generated inputs, generates high-quality 3D assets with rich geometric and textural details. (Project Page: https://magic-research.github.io/magic-boost/)
Authors: Luigi Seminara, Giovanni Maria Farinella, Antonino Furnari
Abstract: Procedural activities are sequences of key-steps aimed at achieving specific goals. They are crucial to build intelligent agents able to assist users effectively. In this context, task graphs have emerged as a human-understandable representation of procedural activities, encoding a partial ordering over the key-steps. While previous works generally relied on hand-crafted procedures to extract task graphs from videos, in this paper, we propose an approach based on direct maximum likelihood optimization of edges' weights, which allows gradient-based learning of task graphs and can be naturally plugged into neural network architectures. Experiments on the CaptainCook4D dataset demonstrate the ability of our approach to predict accurate task graphs from the observation of action sequences, with an improvement of +16.7% over previous approaches. Owing to the differentiability of the proposed framework, we also introduce a feature-based approach, aiming to predict task graphs from key-step textual or video embeddings, for which we observe emerging video understanding abilities. Task graphs learned with our approach are also shown to significantly enhance online mistake detection in procedural egocentric videos, achieving notable gains of +19.8% and +7.5% on the Assembly101-O and EPIC-Tent-O datasets. Code for replicating experiments is available at https://github.com/fpv-iplab/Differentiable-Task-Graph-Learning.
URLs: https://github.com/fpv-iplab/Differentiable-Task-Graph-Learning.
Authors: Faxu Guo, Quan Feng, Sen Yang, Wanxia Yang
Abstract: Hyperspectral remote sensing (HIS) enables the detailed capture of spectral information from the Earth's surface, facilitating precise classification and identification of surface crops due to its superior spectral diagnostic capabilities. However, current convolutional neural networks (CNNs) focus on local features in hyperspectral data, leading to suboptimal performance when classifying intricate crop types and addressing imbalanced sample distributions. In contrast, the Transformer framework excels at extracting global features from hyperspectral imagery. To leverage the strengths of both approaches, this research introduces the Convolutional Meet Transformer Network (CMTNet). This innovative model includes a spectral-spatial feature extraction module for shallow feature capture, a dual-branch structure combining CNN and Transformer branches for local and global feature extraction, and a multi-output constraint module that enhances classification accuracy through multi-output loss calculations and cross constraints across local, international, and joint features. Extensive experiments conducted on three datasets (WHU-Hi-LongKou, WHU-Hi-HanChuan, and WHU-Hi-HongHu) demonstrate that CTDBNet significantly outperforms other state-of-the-art networks in classification performance, validating its effectiveness in hyperspectral crop classification.
Authors: Lucas Mahler, Julius Steiglechner, Benjamin Bender, Tobias Lindig, Dana Ramadan, Jonas Bause, Florian Birk, Rahel Heule, Edyta Charyasz, Michael Erb, Vinod Jangir Kumar, Gisela E Hagberg, Pascal Martin, Gabriele Lohmann, Klaus Scheffler
Abstract: The UltraCortex repository (https://www.ultracortex.org) houses magnetic resonance imaging data of the human brain obtained at an ultra-high field strength of 9.4 T. It contains 86 structural MR images with spatial resolutions ranging from 0.6 to 0.8 mm. Additionally, the repository includes segmentations of 12 brains into gray and white matter compartments. These segmentations have been independently validated by two expert neuroradiologists, thus establishing them as a reliable gold standard. This resource provides researchers with access to high-quality brain imaging data and validated segmentations, facilitating neuroimaging studies and advancing our understanding of brain structure and function. Existing repositories do not accommodate field strengths beyond 7 T, nor do they offer validated segmentations, underscoring the significance of this new resource.
Authors: Ranjan Sapkota, Rizwan Qureshi, Marco Flores Calero, Chetan Badjugar, Upesh Nepal, Alwin Poulose, Peter Zeno, Uday Bhanu Prakash Vaddevolu, Sheheryar Khan, Maged Shoman, Hong Yan, Manoj Karkee
Abstract: Given the rapid emergence and applications of Large Language This review systematically examines the progression of the You Only Look Once (YOLO) object detection algorithms from YOLOv1 to the recently unveiled YOLO11 (or YOLOv11). Employing a reverse chronological analysis, this study examines the advancements introduced by YOLO algorithms, beginning with YOLOv11 and progressing through YOLOv10, YOLOv9, YOLOv8, and subsequent versions to explore each version's contributions to enhancing speed, detection accuracy, and computational efficiency in real-time object detection. By detailing the incremental technological advancements in subsequent YOLO versions, this review chronicles the evolution of YOLO, and discusses the challenges and limitations in each earlier versions. The evolution signifies a path towards integrating YOLO with multimodal, context-aware, and Artificial General Intelligence (AGI) systems for the next YOLO decade, promising significant implications for future developments in AI-driven applications. YOLOV11 to YOLOv1
Authors: Aaron Lohner, Francesco Compagno, Jonathan Francis, Alessandro Oltramari
Abstract: Recognizing a traffic accident is an essential part of any autonomous driving or road monitoring system. An accident can appear in a wide variety of forms, and understanding what type of accident is taking place may be useful to prevent it from recurring. This work focuses on classifying traffic scenes into specific accident types. We approach the problem by representing a traffic scene as a graph, where objects such as cars can be represented as nodes, and relative distances and directions between them as edges. This representation of a traffic scene is referred to as a scene graph, and can be used as input for an accident classifier. Better results are obtained with a classifier that fuses the scene graph input with visual and textual representations. This work introduces a multi-stage, multimodal pipeline that pre-processes videos of traffic accidents, encodes them as scene graphs, and aligns this representation with vision and language modalities before executing the classification task. When trained on 4 classes, our method achieves a balanced accuracy score of 57.77% on an (unbalanced) subset of the popular Detection of Traffic Anomaly (DoTA) benchmark, representing an increase of close to 5 percentage points from the case where scene graph information is not taken into account.
Authors: Duc-Hai Pham, Duc-Dung Nguyen, Anh Pham, Tuan Ho, Phong Nguyen, Khoi Nguyen, Rang Nguyen
Abstract: Accurate prediction of 3D semantic occupancy from 2D visual images is vital in enabling autonomous agents to comprehend their surroundings for planning and navigation. State-of-the-art methods typically employ fully supervised approaches, necessitating a huge labeled dataset acquired through expensive LiDAR sensors and meticulous voxel-wise labeling by human annotators. The resource-intensive nature of this annotating process significantly hampers the application and scalability of these methods. We introduce a novel semi-supervised framework to alleviate the dependency on densely annotated data. Our approach leverages 2D foundation models to generate essential 3D scene geometric and semantic cues, facilitating a more efficient training process. Our framework exhibits notable properties: (1) Generalizability, applicable to various 3D semantic scene completion approaches, including 2D-3D lifting and 3D-2D transformer methods. (2) Effectiveness, as demonstrated through experiments on SemanticKITTI and NYUv2, wherein our method achieves up to 85% of the fully-supervised performance using only 10% labeled data. This approach not only reduces the cost and labor associated with data annotation but also demonstrates the potential for broader adoption in camera-based systems for 3D semantic occupancy prediction.
Authors: Nico Uhlemann, Yipeng Zhou, Tobias Simeon Mohr, Markus Lienkamp
Abstract: This paper explores pedestrian trajectory prediction in urban traffic while focusing on both model accuracy and real-world applicability. While promising approaches exist, they often revolve around pedestrian datasets excluding traffic-related information, or resemble architectures that are either not real-time capable or robust. To address these limitations, we first introduce a dedicated benchmark based on Argoverse 2, specifically targeting pedestrians in traffic environments. Following this, we present Snapshot, a modular, feed-forward neural network that outperforms the current state of the art, reducing the Average Displacement Error (ADE) by 8.8% while utilizing significantly less information. Despite its agent-centric encoding scheme, Snapshot demonstrates scalability, real-time performance, and robustness to varying motion histories. Moreover, by integrating Snapshot into a modular autonomous driving software stack, we showcase its real-world applicability.
Authors: Jiahang Tu, Hao Fu, Fengyu Yang, Hanbin Zhao, Chao Zhang, Hui Qian
Abstract: Tactile sensation plays a crucial role in the development of multi-modal large models and embodied intelligence. To collect tactile data with minimal cost as possible, a series of studies have attempted to generate tactile images by vision-to-touch image translation. However, compared to text modality, visual modality-driven tactile generation cannot accurately depict human tactile sensation. In this work, we analyze the characteristics of tactile images in detail from two granularities: object-level (tactile texture, tactile shape), and sensor-level (gel status). We model these granularities of information through text descriptions and propose a fine-grained Text-to-Touch generation method (TextToucher) to generate high-quality tactile samples. Specifically, we introduce a multimodal large language model to build the text sentences about object-level tactile information and employ a set of learnable text prompts to represent the sensor-level tactile information. To better guide the tactile generation process with the built text information, we fuse the dual grains of text information and explore various dual-grain text conditioning methods within the diffusion transformer architecture. Furthermore, we propose a Contrastive Text-Touch Pre-training (CTTP) metric to precisely evaluate the quality of text-driven generated tactile data. Extensive experiments demonstrate the superiority of our TextToucher method. The source codes will be available at \url{https://github.com/TtuHamg/TextToucher}.
Authors: Daxuan Ren, Hezi Shi, Jianmin Zheng, Jianfei Cai
Abstract: Iso-surface extraction from an implicit field is a fundamental process in various applications of computer vision and graphics. When dealing with geometric shapes with complicated geometric details, many existing algorithms suffer from high computational costs and memory usage. This paper proposes McGrids, a novel approach to improve the efficiency of iso-surface extraction. The key idea is to construct adaptive grids for iso-surface extraction rather than using a simple uniform grid as prior art does. Specifically, we formulate the problem of constructing adaptive grids as a probability sampling problem, which is then solved by Monte Carlo process. We demonstrate McGrids' capability with extensive experiments from both analytical SDFs computed from surface meshes and learned implicit fields from real multiview images. The experiment results show that our McGrids can significantly reduce the number of implicit field queries, resulting in significant memory reduction, while producing high-quality meshes with rich geometric details.
Authors: Chen Hu, Yian Huang, Kexuan Li, Luping Zhang, Chang Long, Yiming Zhu, Tian Pu, Zhenming Peng
Abstract: Infrared small target detection (ISTD) is widely used in civilian and military applications. However, ISTD encounters several challenges, including the tendency for small and dim targets to be obscured by complex backgrounds.To address this issue, we propose the Dynamic Attention Transformer Network (DATransNet), which aims to extract and preserve edge information of small targets.DATransNet employs the Dynamic Attention Transformer (DATrans), simulating central difference convolutions (CDC) to extract and integrate gradient features with deeper features.Furthermore, we propose a global feature extraction module (GFEM) that offers a comprehensive perspective to prevent the network from focusing solely on details while neglecting the background information. We compare the network with state-of-the-art (SOTA) approaches, and the results demonstrate that our method performs effectively. Our source code is available at https://github.com/greekinRoma/DATransNet.
Authors: M. Jehanzeb Mirza, Mengjie Zhao, Zhuoyuan Mao, Sivan Doveh, Wei Lin, Paul Gavrikov, Michael Dorkenwald, Shiqi Yang, Saurav Jha, Hiromi Wakaki, Yuki Mitsufuji, Horst Possegger, Rogerio Feris, Leonid Karlinsky, James Glass
Abstract: In this work, we propose a novel method (GLOV) enabling Large Language Models (LLMs) to act as implicit Optimizers for Vision-Langugage Models (VLMs) to enhance downstream vision tasks. Our GLOV meta-prompts an LLM with the downstream task description, querying it for suitable VLM prompts (e.g., for zero-shot classification with CLIP). These prompts are ranked according to a purity measure obtained through a fitness function. In each respective optimization step, the ranked prompts are fed as in-context examples (with their accuracies) to equip the LLM with the knowledge of the type of text prompts preferred by the downstream VLM. Furthermore, we also explicitly steer the LLM generation process in each optimization step by specifically adding an offset difference vector of the embeddings from the positive and negative solutions found by the LLM, in previous optimization steps, to the intermediate layer of the network for the next generation step. This offset vector steers the LLM generation toward the type of language preferred by the downstream VLM, resulting in enhanced performance on the downstream vision tasks. We comprehensively evaluate our GLOV on 16 diverse datasets using two families of VLMs, i.e., dual-encoder (e.g., CLIP) and encoder-decoder (e.g., LLaVa) models -- showing that the discovered solutions can enhance the recognition performance by up to 15.0% and 57.5% (3.8% and 21.6% on average) for these models.
Authors: Muhammad Awais, Ali Husain Salem Abdulla Alharthi, Amandeep Kumar, Hisham Cholakkal, Rao Muhammad Anwer
Abstract: Significant progress has been made in advancing large multimodal conversational models (LMMs), capitalizing on vast repositories of image-text data available online. Despite this progress, these models often encounter substantial domain gaps, hindering their ability to engage in complex conversations across new domains. Recent efforts have aimed to mitigate this issue, albeit relying on domain-specific image-text data to curate instruction-tuning data. However, many domains, such as agriculture, lack such vision-language data. In this work, we propose an approach to construct instruction-tuning data that harnesses vision-only data for the agriculture domain. We utilize diverse agricultural datasets spanning multiple domains, curate class-specific information, and employ large language models (LLMs) to construct an expert-tuning set, resulting in a 70k expert-tuning dataset called AgroInstruct. Subsequently, we expert-tuned and created AgroGPT, an efficient LMM that can hold complex agriculture-related conversations and provide useful insights. We also develop AgroEvals for evaluation and compare {AgroGPT's} performance with large open and closed-source models. {AgroGPT} excels at identifying fine-grained agricultural concepts, can act as an agriculture expert, and provides helpful information for multimodal agriculture questions. The code, datasets, and models are available at https://github.com/awaisrauf/agroGPT.
Authors: Lihe Yang, Zhen Zhao, Hengshuang Zhao
Abstract: Semi-supervised semantic segmentation (SSS) aims at learning rich visual knowledge from cheap unlabeled images to enhance semantic segmentation capability. Among recent works, UniMatch improves its precedents tremendously by amplifying the practice of weak-to-strong consistency regularization. Subsequent works typically follow similar pipelines and propose various delicate designs. Despite the achieved progress, strangely, even in this flourishing era of numerous powerful vision models, almost all SSS works are still sticking to 1) using outdated ResNet encoders with small-scale ImageNet-1K pre-training, and 2) evaluation on simple Pascal and Cityscapes datasets. In this work, we argue that, it is necessary to switch the baseline of SSS from ResNet-based encoders to more capable ViT-based encoders (e.g., DINOv2) that are pre-trained on massive data. A simple update on the encoder (even using 2x fewer parameters) can bring more significant improvement than careful method designs. Built on this competitive baseline, we present our upgraded and simplified UniMatch V2, inheriting the core spirit of weak-to-strong consistency from V1, but requiring less training cost and providing consistently better results. Additionally, witnessing the gradually saturated performance on Pascal and Cityscapes, we appeal that we should focus on more challenging benchmarks with complex taxonomy, such as ADE20K and COCO datasets. Code, models, and logs of all reported values, are available at https://github.com/LiheYoung/UniMatch-V2.
Authors: Guowei Xu, Peng Jin, Hao Li, Yibing Song, Lichao Sun, Li Yuan
Abstract: Large language models have demonstrated substantial advancements in reasoning capabilities, particularly through inference-time scaling, as illustrated by models such as OpenAI's o1. However, current Vision-Language Models (VLMs) often struggle to perform systematic and structured reasoning, especially when handling complex visual question-answering tasks. In this work, we introduce LLaVA-CoT, a novel VLM designed to conduct autonomous multistage reasoning. Unlike chain-of-thought prompting, LLaVA-CoT independently engages in sequential stages of summarization, visual interpretation, logical reasoning, and conclusion generation. This structured approach enables LLaVA-CoT to achieve marked improvements in precision on reasoning-intensive tasks. To accomplish this, we compile the LLaVA-CoT-100k dataset, integrating samples from various visual question answering sources and providing structured reasoning annotations. Besides, we propose an inference-time stage-level beam search method, which enables effective inference-time scaling. Remarkably, with only 100k training samples and a simple yet effective inference time scaling method, LLaVA-CoT not only outperforms its base model by 7.4% on a wide range of multimodal reasoning benchmarks, but also surpasses the performance of larger and even closed-source models, such as Gemini-1.5-pro, GPT-4o-mini, and Llama-3.2-90B-Vision-Instruct.
Authors: Owen Burns, Rizwan Qureshi
Abstract: We address the issue of the exploding computational requirements of recent State-of-the-art (SOTA) open set multimodel 3D mapping (dense 3D mapping) algorithms and present Voxel-Aggregated Feature Synthesis (VAFS), a novel approach to dense 3D mapping in simulation. Dense 3D mapping involves segmenting and embedding sequential RGBD frames which are then fused into 3D. This leads to redundant computation as the differences between frames are small but all are individually segmented and embedded. This makes dense 3D mapping impractical for research involving embodied agents in which the environment, and thus the mapping, must be modified with regularity. VAFS drastically reduces this computation by using the segmented point cloud computed by a simulator's physics engine and synthesizing views of each region. This reduces the number of features to embed from the number of captured RGBD frames to the number of objects in the scene, effectively allowing a "ground truth" semantic map to be computed an order of magnitude faster than traditional methods. We test the resulting representation by assessing the IoU scores of semantic queries for different objects in the simulated scene, and find that VAFS exceeds the accuracy and speed of prior dense 3D mapping techniques.
Authors: Shahriar Soudeep, M. F. Mridha, Md Abrar Jahin, Nilanjan Dey
Abstract: The detection and tracking of small, occluded objects such as pedestrians, cyclists, and motorbikes pose significant challenges for traffic surveillance systems because of their erratic movement, frequent occlusion, and poor visibility in dynamic urban environments. Traditional methods like YOLO11, while proficient in spatial feature extraction for precise detection, often struggle with these small and dynamically moving objects, particularly in handling real-time data updates and resource efficiency. This paper introduces DGNN-YOLO, a novel framework that integrates dynamic graph neural networks (DGNNs) with YOLO11 to address these limitations. Unlike standard GNNs, DGNNs are chosen for their superior ability to dynamically update graph structures in real-time, which enables adaptive and robust tracking of objects in highly variable urban traffic scenarios. This framework constructs and regularly updates its graph representations, capturing objects as nodes and their interactions as edges, thus effectively responding to rapidly changing conditions. Additionally, DGNN-YOLO incorporates Grad-CAM, Grad-CAM++, and Eigen-CAM visualization techniques to enhance interpretability and foster trust, offering insights into the model's decision-making process. Extensive experiments validate the framework's performance, achieving a precision of 0.8382, recall of 0.6875, and mAP@0.5:0.95 of 0.6476, significantly outperforming existing methods. This study offers a scalable and interpretable solution for real-time traffic surveillance and significantly advances intelligent transportation systems' capabilities by addressing the critical challenge of detecting and tracking small, occluded objects.
Authors: Huajian Zeng, Maolin Gao, Daniel Cremers
Abstract: The interest in matching non-rigidly deformed shapes represented as raw point clouds is rising due to the proliferation of low-cost 3D sensors. Yet, the task is challenging since point clouds are irregular and there is a lack of intrinsic shape information. We propose to tackle these challenges by learning a new shape representation -- a per-point high dimensional embedding, in an embedding space where semantically similar points share similar embeddings. The learned embedding has multiple beneficial properties: it is aware of the underlying shape geometry and is robust to shape deformations and various shape artefacts, such as noise and partiality. Consequently, this embedding can be directly employed to retrieve high-quality dense correspondences through a simple nearest neighbor search in the embedding space. Extensive experiments demonstrate new state-of-the-art results and robustness in numerous challenging non-rigid shape matching benchmarks and show its great potential in other shape analysis tasks, such as segmentation.
Authors: Junjie Wang, Yuze Gao, Dongying Li, Wenxian Yu
Abstract: Detecting small targets in sea clutter is challenging due to dynamic maritime conditions. Existing solutions either model sea clutter for detection or extract target features based on clutter-target echo differences, including statistical and deep features. While more common, the latter often excels in controlled scenarios but struggles with robust detection and generalization in diverse environments, limiting practical use. In this letter, we propose a multi-domain features guided supervised contrastive learning (MDFG_SCL) method, which integrates statistical features derived from multi-domain differences with deep features obtained through supervised contrastive learning, thereby capturing both low-level domain-specific variations and high-level semantic information. This comprehensive feature integration enables the model to effectively distinguish between small targets and sea clutter, even under challenging conditions. Experiments conducted on real-world datasets demonstrate that the proposed shallow-to-deep detector not only achieves effective identification of small maritime targets but also maintains superior detection performance across varying sea conditions, outperforming the mainstream unsupervised contrastive learning and supervised contrastive learning methods.
Authors: Anushrut Jignasu, Ethan Herron, Zhanhong Jiang, Soumik Sarkar, Chinmay Hegde, Baskar Ganapathysubramanian, Aditya Balu, Adarsh Krishnamurthy
Abstract: We present STITCH, a novel approach for neural implicit surface reconstruction of a sparse and irregularly spaced point cloud while enforcing topological constraints (such as having a single connected component). We develop a new differentiable framework based on persistent homology to formulate topological loss terms that enforce the prior of a single 2-manifold object. Our method demonstrates excellent performance in preserving the topology of complex 3D geometries, evident through both visual and empirical comparisons. We supplement this with a theoretical analysis, and provably show that optimizing the loss with stochastic (sub)gradient descent leads to convergence and enables reconstructing shapes with a single connected component. Our approach showcases the integration of differentiable topological data analysis tools for implicit surface reconstruction.
Authors: Yue Fan, Xiaojian Ma, Rongpeng Su, Jun Guo, Rujie Wu, Xi Chen, Qing Li
Abstract: This paper investigates the problem of understanding dynamic 3D scenes from egocentric observations, a key challenge in robotics and embodied AI. Unlike prior studies that explored this as long-form video understanding and utilized egocentric video only, we instead propose an LLM-based agent, Embodied VideoAgent, which constructs scene memory from both egocentric video and embodied sensory inputs (e.g. depth and pose sensing). We further introduce a VLM-based approach to automatically update the memory when actions or activities over objects are perceived. Embodied VideoAgent attains significant advantages over counterparts in challenging reasoning and planning tasks in 3D scenes, achieving gains of 4.9% on Ego4D-VQ3D, 5.8% on OpenEQA, and 11.7% on EnvQA. We have also demonstrated its potential in various embodied AI tasks including generating embodied interactions and perception for robot manipulation. The code and demo will be made public.
Authors: Zhangyang Qi, Zhixiong Zhang, Ye Fang, Jiaqi Wang, Hengshuang Zhao
Abstract: In recent years, 2D Vision-Language Models (VLMs) have made significant strides in image-text understanding tasks. However, their performance in 3D spatial comprehension, which is critical for embodied intelligence, remains limited. Recent advances have leveraged 3D point clouds and multi-view images as inputs, yielding promising results. However, we propose exploring a purely vision-based solution inspired by human perception, which merely relies on visual cues for 3D spatial understanding. This paper empirically investigates the limitations of VLMs in 3D spatial knowledge, revealing that their primary shortcoming lies in the lack of global-local correspondence between the scene and individual frames. To address this, we introduce GPT4Scene, a novel visual prompting paradigm in VLM training and inference that helps build the global-local relationship, significantly improving the 3D spatial understanding of indoor scenes. Specifically, GPT4Scene constructs a 3D Bird's Eye View (BEV) image from the video and marks consistent object IDs across both frames and the BEV image. The model then inputs the concatenated BEV image and video frames with markers. In zero-shot evaluations, GPT4Scene improves performance over closed-source VLMs like GPT-4o. Additionally, we prepare a processed video dataset consisting of 165K text annotation to fine-tune open-source VLMs, achieving state-of-the-art performance on all 3D understanding tasks. Surprisingly, after training with the GPT4Scene paradigm, VLMs consistently improve during inference, even without visual prompting and BEV image as explicit correspondence. It demonstrates that the proposed paradigm helps VLMs develop an intrinsic ability to understand 3D scenes, which paves the way for a noninvasive approach to extending pre-trained VLMs for 3D scene understanding.
Authors: Huaize Liu, Wenzhang Sun, Donglin Di, Shibo Sun, Jiahui Yang, Changqing Zou, Hujun Bao
Abstract: The generation of talking avatars has achieved significant advancements in precise audio synchronization. However, crafting lifelike talking head videos requires capturing a broad spectrum of emotions and subtle facial expressions. Current methods face fundamental challenges: a) the absence of frameworks for modeling single basic emotional expressions, which restricts the generation of complex emotions such as compound emotions; b) the lack of comprehensive datasets rich in human emotional expressions, which limits the potential of models. To address these challenges, we propose the following innovations: 1) the Mixture of Emotion Experts (MoEE) model, which decouples six fundamental emotions to enable the precise synthesis of both singular and compound emotional states; 2) the DH-FaceEmoVid-150 dataset, specifically curated to include six prevalent human emotional expressions as well as four types of compound emotions, thereby expanding the training potential of emotion-driven models. Furthermore, to enhance the flexibility of emotion control, we propose an emotion-to-latents module that leverages multimodal inputs, aligning diverse control signals-such as audio, text, and labels-to ensure more varied control inputs as well as the ability to control emotions using audio alone. Through extensive quantitative and qualitative evaluations, we demonstrate that the MoEE framework, in conjunction with the DH-FaceEmoVid-150 dataset, excels in generating complex emotional expressions and nuanced facial details, setting a new benchmark in the field. These datasets will be publicly released.
Authors: Di Jin, Xing Liu, Yu Liu, Jia Qing Yap, Andrea Wong, Adriana Crespo, Qi Lin, Zhiyuan Yin, Qiang Yan, Ryan Ye
Abstract: The rapid development of large language models (LLMs) and large vision models (LVMs) have propelled the evolution of multi-modal AI systems, which have demonstrated the remarkable potential for industrial applications by emulating human-like cognition. However, they also pose significant ethical challenges, including amplifying harmful content and reinforcing societal biases. For instance, biases in some industrial image generation models highlighted the urgent need for robust fairness assessments. Most existing evaluation frameworks focus on the comprehensiveness of various aspects of the models, but they exhibit critical limitations, including insufficient attention to content generation alignment and social bias-sensitive domains. More importantly, their reliance on pixel-detection techniques is prone to inaccuracies. To address these issues, this paper presents INFELM, an in-depth fairness evaluation on widely-used text-to-image models. Our key contributions are: (1) an advanced skintone classifier incorporating facial topology and refined skin pixel representation to enhance classification precision by at least 16.04%, (2) a bias-sensitive content alignment measurement for understanding societal impacts, (3) a generalizable representation bias evaluation for diverse demographic groups, and (4) extensive experiments analyzing large-scale text-to-image model outputs across six social-bias-sensitive domains. We find that existing models in the study generally do not meet the empirical fairness criteria, and representation bias is generally more pronounced than alignment errors. INFELM establishes a robust benchmark for fairness assessment, supporting the development of multi-modal AI systems that align with ethical and human-centric principles.
Authors: Mengting Wei, Tuomas Varanka, Xingxun Jiang, Huai-Qian Khor, Guoying Zhao
Abstract: We address the problem of facial expression editing by controling the relative variation of facial action-unit (AU) from the same person. This enables us to edit this specific person's expression in a fine-grained, continuous and interpretable manner, while preserving their identity, pose, background and detailed facial attributes. Key to our model, which we dub MagicFace, is a diffusion model conditioned on AU variations and an ID encoder to preserve facial details of high consistency. Specifically, to preserve the facial details with the input identity, we leverage the power of pretrained Stable-Diffusion models and design an ID encoder to merge appearance features through self-attention. To keep background and pose consistency, we introduce an efficient Attribute Controller by explicitly informing the model of current background and pose of the target. By injecting AU variations into a denoising UNet, our model can animate arbitrary identities with various AU combinations, yielding superior results in high-fidelity expression editing compared to other facial expression editing works. Code is publicly available at https://github.com/weimengting/MagicFace.
Authors: Chan Lee, Seungho Shin, Gyeong-Moon Park, Jung Uk Kim
Abstract: Although existing Sparsely Annotated Object Detection (SAOD) approches have made progress in handling sparsely annotated environments in multispectral domain, where only some pedestrians are annotated, they still have the following limitations: (i) they lack considerations for improving the quality of pseudo-labels for missing annotations, and (ii) they rely on fixed ground truth annotations, which leads to learning only a limited range of pedestrian visual appearances in the multispectral domain. To address these issues, we propose a novel framework called Sparsely Annotated Multispectral Pedestrian Detection (SAMPD). For limitation (i), we introduce Multispectral Pedestrian-aware Adaptive Weight (MPAW) and Positive Pseudo-label Enhancement (PPE) module. Utilizing multispectral knowledge, these modules ensure the generation of high-quality pseudo-labels and enable effective learning by increasing weights for high-quality pseudo-labels based on modality characteristics. To address limitation (ii), we propose an Adaptive Pedestrian Retrieval Augmentation (APRA) module, which adaptively incorporates pedestrian patches from ground-truth and dynamically integrates high-quality pseudo-labels with the ground-truth, facilitating a more diverse learning pool of pedestrians. Extensive experimental results demonstrate that our SAMPD significantly enhances performance in sparsely annotated environments within the multispectral domain.
Authors: Valery Istomin, Oleg Pereziabov, Ilya Afanasyev
Abstract: This research focuses on developing a method for restoring the topology of digital images of paper documents captured by a camera, using algorithms for detection, segmentation, geometry restoration, and dewarping. Our methodology employs deep learning (DL) for document outline detection, followed by computer vision (CV) to create a topological 2D grid using cubic polynomial interpolation and correct nonlinear distortions by remapping the image. Using classical CV methods makes the document topology restoration process more efficient and faster, as it requires significantly fewer computational resources and memory. We developed a new pipeline for automatic document dewarping and reconstruction, along with a framework and annotated dataset to demonstrate its efficiency. Our experiments confirm the promise of our methodology and its superiority over existing benchmarks (including mobile apps and popular DL solutions, such as RectiNet, DocGeoNet, and DocTr++) both visually and in terms of document readability via Optical Character Recognition (OCR) and geometry restoration metrics. This paves the way for creating high-quality digital copies of paper documents and enhancing the efficiency of OCR systems. Project page: https://github.com/HorizonParadox/DRCCBI
Authors: Xuyang Wang, Ziang Cheng, Zhenyu Li, Jiayu Yang, Haorui Ji, Pan Ji, Mehrtash Harandi, Richard Hartley, Hongdong Li
Abstract: This paper proposes DoubleDiffusion, a novel framework that combines heat dissipation diffusion and denoising diffusion for direct generative learning on 3D mesh surfaces. Our approach addresses the challenges of generating continuous signal distributions residing on a curve manifold surface. Unlike previous methods that rely on unrolling 3D meshes into 2D or adopting field representations, DoubleDiffusion leverages the Laplacian-Beltrami operator to process features respecting the mesh structure. This combination enables effective geometry-aware signal diffusion across the underlying geometry. As shown in Fig.1, we demonstrate that DoubleDiffusion has the ability to generate RGB signal distributions on complex 3D mesh surfaces and achieves per-category shape-conditioned texture generation across different shape geometry. Our work contributes a new direction in diffusion-based generative modeling on 3D surfaces, with potential applications in the field of 3D asset generation.
Authors: Zhongxuan Zhang, Bi Zeng, Xinyu Ni, Yimin Du
Abstract: RGB-T tracking leverages the complementary strengths of RGB and thermal infrared (TIR) modalities to address challenging scenarios such as low illumination and adverse weather. However, existing methods often fail to effectively integrate temporal information and perform efficient cross-modal interactions, which constrain their adaptability to dynamic targets. In this paper, we propose BTMTrack, a novel framework for RGB-T tracking. The core of our approach lies in the dual-template backbone network and the Temporal-Modal Candidate Elimination (TMCE) strategy. The dual-template backbone effectively integrates temporal information, while the TMCE strategy focuses the model on target-relevant tokens by evaluating temporal and modal correlations, reducing computational overhead and avoiding irrelevant background noise. Building upon this foundation, we propose the Temporal Dual Template Bridging (TDTB) module, which facilitates precise cross-modal fusion through dynamically filtered tokens. This approach further strengthens the interaction between templates and the search region. Extensive experiments conducted on three benchmark datasets demonstrate the effectiveness of BTMTrack. Our method achieves state-of-the-art performance, with a 72.3% precision rate on the LasHeR test set and competitive results on RGBT210 and RGBT234 datasets.
Authors: Zekai Gu, Rui Yan, Jiahao Lu, Peng Li, Zhiyang Dou, Chenyang Si, Zhen Dong, Qifeng Liu, Cheng Lin, Ziwei Liu, Wenping Wang, Yuan Liu
Abstract: Diffusion models have demonstrated impressive performance in generating high-quality videos from text prompts or images. However, precise control over the video generation process, such as camera manipulation or content editing, remains a significant challenge. Existing methods for controlled video generation are typically limited to a single control type, lacking the flexibility to handle diverse control demands. In this paper, we introduce Diffusion as Shader (DaS), a novel approach that supports multiple video control tasks within a unified architecture. Our key insight is that achieving versatile video control necessitates leveraging 3D control signals, as videos are fundamentally 2D renderings of dynamic 3D content. Unlike prior methods limited to 2D control signals, DaS leverages 3D tracking videos as control inputs, making the video diffusion process inherently 3D-aware. This innovation allows DaS to achieve a wide range of video controls by simply manipulating the 3D tracking videos. A further advantage of using 3D tracking videos is their ability to effectively link frames, significantly enhancing the temporal consistency of the generated videos. With just 3 days of fine-tuning on 8 H800 GPUs using less than 10k videos, DaS demonstrates strong control capabilities across diverse tasks, including mesh-to-video generation, camera control, motion transfer, and object manipulation.
Authors: Hyungjin Chung, Dohun Lee, Zihui Wu, Byung-Hoon Kim, Katherine L. Bouman, Jong Chul Ye
Abstract: Compressed sensing MRI seeks to accelerate MRI acquisition processes by sampling fewer k-space measurements and then reconstructing the missing data algorithmically. The success of these approaches often relies on strong priors or learned statistical models. While recent diffusion model-based priors have shown great potential, previous methods typically ignore clinically available metadata (e.g. patient demographics, imaging parameters, slice-specific information). In practice, metadata contains meaningful cues about the anatomy and acquisition protocol, suggesting it could further constrain the reconstruction problem. In this work, we propose ContextMRI, a text-conditioned diffusion model for MRI that integrates granular metadata into the reconstruction process. We train a pixel-space diffusion model directly on minimally processed, complex-valued MRI images. During inference, metadata is converted into a structured text prompt and fed to the model via CLIP text embeddings. By conditioning the prior on metadata, we unlock more accurate reconstructions and show consistent gains across multiple datasets, acceleration factors, and undersampling patterns. Our experiments demonstrate that increasing the fidelity of metadata, ranging from slice location and contrast to patient age, sex, and pathology, systematically boosts reconstruction performance. This work highlights the untapped potential of leveraging clinical context for inverse problems and opens a new direction for metadata-driven MRI reconstruction.
Authors: Runzhen Liu, Qinjie Lin, Yunfei Liu, Lijian Lin, Ye Zhu, Yu Li, Chuhua Xian, Fa-Ting Hong
Abstract: Video dubbing aims to synthesize realistic, lip-synced videos from a reference video and a driving audio signal. Although existing methods can accurately generate mouth shapes driven by audio, they often fail to preserve identity-specific features, largely because they do not effectively capture the nuanced interplay between audio cues and the visual attributes of reference identity . As a result, the generated outputs frequently lack fidelity in reproducing the unique textural and structural details of the reference identity. To address these limitations, we propose IPTalker, a novel and robust framework for video dubbing that achieves seamless alignment between driving audio and reference identity while ensuring both lip-sync accuracy and high-fidelity identity preservation. At the core of IPTalker is a transformer-based alignment mechanism designed to dynamically capture and model the correspondence between audio features and reference images, thereby enabling precise, identity-aware audio-visual integration. Building on this alignment, a motion warping strategy further refines the results by spatially deforming reference images to match the target audio-driven configuration. A dedicated refinement process then mitigates occlusion artifacts and enhances the preservation of fine-grained textures, such as mouth details and skin features. Extensive qualitative and quantitative evaluations demonstrate that IPTalker consistently outperforms existing approaches in terms of realism, lip synchronization, and identity retention, establishing a new state of the art for high-quality, identity-consistent video dubbing.
Authors: Carlos Guti\'errez-\'Alvarez, Pablo R\'ios-Navarro, Rafael Flor-Rodr\'iguez, Francisco Javier Acevedo-Rodr\'iguez, Roberto J. L\'opez-Sastre
Abstract: Visual Semantic Navigation (VSN) is the ability of a robot to learn visual semantic information for navigating in unseen environments. These VSN models are typically tested in those virtual environments where they are trained, mainly using reinforcement learning based approaches. Therefore, we do not yet have an in-depth analysis of how these models would behave in the real world. In this work, we propose a new solution to integrate VSN models into real robots, so that we have true embodied agents. We also release a novel ROS-based framework for VSN, ROS4VSN, so that any VSN-model can be easily deployed in any ROS-compatible robot and tested in a real setting. Our experiments with two different robots, where we have embedded two state-of-the-art VSN agents, confirm that there is a noticeable performance difference of these VSN solutions when tested in real-world and simulation environments. We hope that this research will endeavor to provide a foundation for addressing this consequential issue, with the ultimate aim of advancing the performance and efficiency of embodied agents within authentic real-world scenarios. Code to reproduce all our experiments can be found at https://github.com/gramuah/ros4vsn.
Authors: Jianeng Wang, Matias Mattamala, Christina Kassab, Guillaume Burger, Fabio Elnecave, Lintong Zhang, Marine Petriaux, Maurice Fallon
Abstract: Self-balancing exoskeletons are a key enabling technology for individuals with mobility impairments. While the current challenges focus on human-compliant hardware and control, unlocking their use for daily activities requires a scene perception system. In this work, we present Exosense, a vision-centric scene understanding system for self-balancing exoskeletons. We introduce a multi-sensor visual-inertial mapping device as well as a navigation stack for state estimation, terrain mapping and long-term operation. We tested Exosense attached to both a human leg and Wandercraft's Personal Exoskeleton in real-world indoor scenarios. This enabled us to test the system during typical periodic walking gaits, as well as future uses in multi-story environments. We demonstrate that Exosense can achieve an odometry drift of about 4 cm per meter traveled, and construct terrain maps under 1 cm average reconstruction error. It can also work in a visual localization mode in a previously mapped environment, providing a step towards long-term operation of exoskeletons.
Authors: Pengcheng Chen, Wenhao Li, Nicole Gunderson, Jeremy Ruthberg, Randall Bly, Zhenglong Sun, Waleed M. Abuzeid, Eric J. Seibel
Abstract: 3D reconstruction in endoscopic sinus surgery (ESS) demands exceptional accuracy, with the mean error and standard deviation necessitating within the range of a single CT slice (0.625 mm), as the critical structures in the nasal cavity are situated within submillimeter distances from surgical instruments. This poses a formidable challenge when using conventional monocular endoscopes. Depth estimation is crucial for 3D reconstruction, yet existing depth estimation methodologies either suffer from inherent accuracy limitations or, in the case of learning-based approaches, perform poorly when applied to ESS despite succeeding on their original datasets. In this study, we present a novel, highly generalizable method that combines Neural Radiance Fields (NeRF) and stereo depth estimation for 3D reconstruction that can derive metric monocular depth. Our approach begins with an initial NeRF reconstruction yielding a coarse 3D scene, the subsequent creation of binocular pairs within coarse 3D scene, and generation of depth maps through stereo vision, These depth maps are used to supervise subsequent NeRF iteration, progressively refining NeRF and binocular depth, the refinement process continues until the depth maps converged. This recursive process generates high-accuracy depth maps from monocular endoscopic video. Evaluation in synthetic endoscopy shows a depth accuracy of 0.125 $\pm$ 0.443 mm, well within the 0.625 mm threshold. Further clinical experiments with real endoscopic data demonstrate a mean distance to CT mesh of 0.269 mm, representing the highest accuracy among monocular 3D reconstruction methods in ESS.
Authors: Xiao Huo, Junhui Hou, Shuai Wan, Fuzheng Yang
Abstract: The evolution of 3D visualization techniques has fundamentally transformed how we interact with digital content. At the forefront of this change is point cloud technology, offering an immersive experience that surpasses traditional 2D representations. However, the massive data size of point clouds presents significant challenges in data compression. Current methods for lossy point cloud attribute compression (PCAC) generally focus on reconstructing the original point clouds with minimal error. However, for point cloud visualization scenarios, the reconstructed point clouds with distortion still need to undergo a complex rendering process, which affects the final user-perceived quality. In this paper, we propose an end-to-end deep learning framework that seamlessly integrates PCAC with differentiable rendering, denoted as rendering-oriented PCAC (RO-PCAC), directly targeting the quality of rendered multiview images for viewing. In a differentiable manner, the impact of the rendering process on the reconstructed point clouds is taken into account. Moreover, we characterize point clouds as sparse tensors and propose a sparse tensor-based transformer, called SP-Trans. By aligning with the local density of the point cloud and utilizing an enhanced local attention mechanism, SP-Trans captures the intricate relationships within the point cloud, further improving feature analysis and synthesis within the framework. Extensive experiments demonstrate that the proposed RO-PCAC achieves state-of-the-art compression performance, compared to existing reconstruction-oriented methods, including traditional, learning-based, and hybrid methods.
Authors: Feng Xiong, Runxi Cheng, Wang Chen, Zhanqiu Zhang, Yiwen Guo, Chun Yuan, Ruifeng Xu
Abstract: Model merging has recently gained attention as an economical and scalable approach to incorporate task-specific weights from various tasks into a unified multi-task model. For example, in Task Arithmetic (TA), adding the fine-tuned weights of different tasks can enhance the model's performance on those tasks, while subtracting them leads to task forgetting. Although TA is highly effective, interference among task still hampers the performance of the merged model. Existing methods for handling conflicts between task generally rely on empirical selection, resulting in suboptimal performance. In this paper, we introduce an Adaptive Weight Disentanglement method. We begin by theoretically proving that task vectors employed in model merging should be orthogonal to minimize interference among tasks. Guided by this insight, we initialize redundant vectors such that, when subtracted from the original task vectors, the resulting vectors exhibit increased orthogonality. Additionally, we impose an norm constraint on the redundant vectors to preserve the performance of the task-specific models. Experimental results demonstrate the effectiveness of our proposed technique: it successfully extracts redundant vectors, and after their subtraction, the task vectors not only retain robust performance but also achieve superior fusion outcomes. Our code is available at \href{https://github.com/FarisXiong/AWD.git}{https://github.com/FarisXiong/AWD.git}.
URLs: https://github.com/FarisXiong/AWD.git, https://github.com/FarisXiong/AWD.git
Authors: Ankit Kumar Patel, Dewanshi Paul, Sarthak Giri, Sneha Chaudhary, Bikalpa Gautam
Abstract: Security systems relying on passwords are vulnerable to being forgotten, guessed, or breached. Likewise, biometric systems that operate independently are at risk of template spoofing and replay incidents. This paper introduces a biocryptosystem utilizing face recognition techniques to address these issues, allowing for the encryption and decryption of various file types through the Advanced Encryption Standard (AES). The proposed system creates a distinct 32-bit encryption key derived from facial features identified by Histogram of Oriented Gradients (HOG) and categorized using Support Vector Machines (SVM). HOG efficiently identifies edge-aligned facial features, even in dim lighting, ensuring that reliable biometric keys can be generated. This key is then used with AES to encrypt and decrypt a variety of data formats, such as text, audio, and video files. This encryption key, derived from an individual's distinctive facial traits, is exceedingly challenging for adversaries to reproduce or guess. The security and performance of the system have been validated through experiments using several metrics, including correlation analysis, Shannon entropy, normalized Hamming distance, and the avalanche effect on 25 different file types. Potential uses for the proposed system include secure file sharing, online transactions, and data archiving, making it a strong and trustworthy approach to safeguarding sensitive information by integrating the uniqueness of facial biometrics with the established security of AES encryption.
Authors: Guanghua He, Wangang Cheng, Hancan Zhu, Xiaohao Cai, Gaohang Yu
Abstract: Transfer learning, by leveraging knowledge from pre-trained models, has significantly enhanced the performance of target tasks. However, as deep neural networks scale up, full fine-tuning introduces substantial computational and storage challenges in resource-constrained environments, limiting its widespread adoption. To address this, parameter-efficient fine-tuning (PEFT) methods have been developed to reduce computational complexity and storage requirements by minimizing the number of updated parameters. While matrix decomposition-based PEFT methods, such as LoRA, show promise, they struggle to fully capture the high-dimensional structural characteristics of model weights. In contrast, high-dimensional tensors offer a more natural representation of neural network weights, allowing for a more comprehensive capture of higher-order features and multi-dimensional interactions. In this paper, we propose tCURLoRA, a novel fine-tuning method based on tensor CUR decomposition. By concatenating pre-trained weight matrices into a three-dimensional tensor and applying tensor CUR decomposition, we update only the lower-order tensor components during fine-tuning, effectively reducing computational and storage overhead. Experimental results demonstrate that tCURLoRA outperforms existing PEFT methods in medical image segmentation tasks.
Authors: Zhaoyi Yan, Zhijie Sang, Yiming Zhang, Yuhao Fu, Baoyi He, Qi Zhou, Yining Di, Chunlin Ji, Shengyu Zhang, Fei Wu, Hongxia Yang
Abstract: Large Language Models (LLMs) have demonstrated strong performance across various reasoning tasks, yet building a single model that consistently excels across all domains remains challenging. This paper addresses this problem by exploring strategies to integrate multiple domain-specialized models into an efficient pivot model.We propose two fusion strategies to combine the strengths of multiple LLMs: (1) a pairwise, multi-step fusion approach that sequentially distills each source model into the pivot model, followed by a weight merging step to integrate the distilled models into the final model. This method achieves strong performance but requires substantial training effort; and (2) a unified fusion approach that aggregates all source models' outputs simultaneously.To improve the fusion process, we introduce a novel Rate-Skewness Adaptive Fusion (RSAF) technique, which dynamically adjusts top-K ratios during parameter merging for enhanced flexibility and stability.Furthermore, we propose an uncertainty-based weighting method for the unified approach, which dynamically balances the contributions of source models and outperforms other logits/distribution ensemble methods.We achieved accuracy improvements of 9.27%, 8.80%, and 8.89% on the GSM8K, MATH, and HumanEval tasks, respectively.
Authors: Run Luo, Ting-En Lin, Haonan Zhang, Yuchuan Wu, Xiong Liu, Min Yang, Yongbin Li, Longze Chen, Jiaming Li, Lei Zhang, Yangyi Chen, Hamid Alinejad-Rokny, Fei Huang
Abstract: Recent advancements in omnimodal learning have been achieved in understanding and generation across images, text, and speech, though mainly within proprietary models. Limited omnimodal datasets and the inherent challenges associated with real-time emotional speech generation have hindered open-source progress. To address these issues, we propose openomni, a two-stage training method combining omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model is further trained on text-image tasks to generalize from vision to speech in a (near) zero-shot manner, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder facilitates real-time emotional speech through training on speech tasks and preference learning. Experiments demonstrate that openomni consistently improves across omnimodal, vision-language, and speech-language evaluations, enabling natural, emotion-rich dialogues and real-time emotional speech generation.