Authors: Hamza Rawal, Muhammad Junaid Ahmad, Farooq Zaman
Flow based garment warping is an integral part of image-based virtual try-on networks. However, optimizing a single flow predicting network for simultaneous global boundary alignment and local texture preservation results in sub-optimal flow fields. Moreover, dense flows are inherently not suited to handle intricate conditions like garment occlusion by body parts or by other garments. Forcing flows to handle the above issues results in various distortions like texture squeezing, and stretching. In this work, we propose a novel approach where we disentangle the global boundary alignment and local texture preserving tasks via our GlobalNet and LocalNet modules. A consistency loss is then employed between the two modules which harmonizes the local flows with the global boundary alignment. Additionally, we explicitly handle occlusions by predicting body-parts visibility mask, which is used to mask out the occluded regions in the warped garment. The masking prevents the LocalNet from predicting flows that distort texture to compensate for occlusions. We also introduce a novel regularization loss (NIPR), that defines a criteria to identify the regions in the warped garment where texture integrity is violated (squeezed or stretched). NIPR subsequently penalizes the flow in those regions to ensure regular and coherent warps that preserve the texture in local neighborhoods. Evaluation on a widely used virtual try-on dataset demonstrates strong performance of our network compared to the current SOTA methods.
Authors: Prasad Gabbur
We propose using a Gaussian Mixture Model (GMM) as reverse transition operator (kernel) within the Denoising Diffusion Implicit Models (DDIM) framework, which is one of the most widely used approaches for accelerated sampling from pre-trained Denoising Diffusion Probabilistic Models (DDPM). Specifically we match the first and second order central moments of the DDPM forward marginals by constraining the parameters of the GMM. We see that moment matching is sufficient to obtain samples with equal or better quality than the original DDIM with Gaussian kernels. We provide experimental results with unconditional models trained on CelebAHQ and FFHQ and class-conditional models trained on ImageNet datasets respectively. Our results suggest that using the GMM kernel leads to significant improvements in the quality of the generated samples when the number of sampling steps is small, as measured by FID and IS metrics. For example on ImageNet 256x256, using 10 sampling steps, we achieve a FID of 6.94 and IS of 207.85 with a GMM kernel compared to 10.15 and 196.73 respectively with a Gaussian kernel.
Authors: Jin-Jian Xu, Hao Zhang, Chao-Sheng Tang, Lin Li, Bin Shi
As Earth science enters the era of big data, artificial intelligence (AI) not only offers great potential for solving geoscience problems, but also plays a critical role in accelerating the understanding of the complex, interactive, and multiscale processes of Earth's behavior. As geoscience AI models are progressively utilized for significant predictions in crucial situations, geoscience researchers are increasingly demanding their interpretability and versatility. This study proposes an interpretable geoscience artificial intelligence (XGeoS-AI) framework to unravel the mystery of image recognition in the Earth sciences, and its effectiveness and versatility is demonstrated by taking computed tomography (CT) image recognition as an example. Inspired by the mechanism of human vision, the proposed XGeoS-AI framework generates a threshold value from a local region within the whole image to complete the recognition. Different kinds of artificial intelligence (AI) methods, such as Support Vector Regression (SVR), Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), can be adopted as the AI engines of the proposed XGeoS-AI framework to efficiently complete geoscience image recognition tasks. Experimental results demonstrate that the effectiveness, versatility, and heuristics of the proposed framework have great potential in solving geoscience image recognition problems. Interpretable AI should receive more and more attention in the field of the Earth sciences, which is the key to promoting more rational and wider applications of AI in the field of Earth sciences. In addition, the proposed interpretable framework may be the forerunner of technological innovation in the Earth sciences.
Authors: Alex Ling Yu Hung, Haoxin Zheng, Kai Zhao, Xiaoxi Du, Kaifeng Pang, Qi Miao, Steven S. Raman, Demetri Terzopoulos, Kyunghyun Sung
A large portion of volumetric medical data, especially magnetic resonance imaging (MRI) data, is anisotropic, as the through-plane resolution is typically much lower than the in-plane resolution. Both 3D and purely 2D deep learning-based segmentation methods are deficient in dealing with such volumetric data since the performance of 3D methods suffers when confronting anisotropic data, and 2D methods disregard crucial volumetric information. Insufficient work has been done on 2.5D methods, in which 2D convolution is mainly used in concert with volumetric information. These models focus on learning the relationship across slices, but typically have many parameters to train. We offer a Cross-Slice Attention Module (CSAM) with minimal trainable parameters, which captures information across all the slices in the volume by applying semantic, positional, and slice attention on deep feature maps at different scales. Our extensive experiments using different network architectures and tasks demonstrate the usefulness and generalizability of CSAM. Associated code is available at https://github.com/aL3x-O-o-Hung/CSAM.
Authors: Siao Tang, Xin Wang, Hong Chen, Chaoyu Guan, Yansong Tang, Wenwu zhu
Diffusion models have recently shown remarkable generation ability, achieving state-of-the-art performance in many tasks. However, the high computational cost is still a troubling problem for diffusion models. To tackle this problem, we propose to automatically remove the structural redundancy in diffusion models with our proposed Diffusion Distillation-based Block-wise Neural Architecture Search (DiffNAS). Specifically, given a larger pretrained teacher, we leverage DiffNAS to search for the smallest architecture which achieves on-par or even better performance than the teacher. Considering current diffusion models are based on UNet which naturally has a block-wise structure, we perform neural architecture search independently in each block, which largely reduces the search space. Different from previous block-wise NAS methods, DiffNAS contains a block-wise local search strategy and a retraining strategy with a joint dynamic loss. Concretely, during the search process, we block-wisely select the best subnet to avoid the unfairness brought by the global search strategy used in previous works. When retraining the searched architecture, we adopt a dynamic joint loss to maintain the consistency between supernet training and subnet retraining, which also provides informative objectives for each block and shortens the paths of gradient propagation. We demonstrate this joint loss can effectively improve model performance. We also prove the necessity of the dynamic adjustment of this loss. The experiments show that our method can achieve significant computational reduction, especially on latent diffusion models with about 50% MACs and Parameter reduction.
Authors: Johnathan Chiu, Samuel Duffield, Max Hunter-Gordon, Kaelan Donatella, Max Aifer, Andi Gu
We introduce a novel approach to video modeling that leverages controlled differential equations (CDEs) to address key challenges in video tasks, notably video interpolation and mask propagation. We apply CDEs at varying resolutions leading to a continuous-time U-Net architecture. Unlike traditional methods, our approach does not require explicit optical flow learning, and instead makes use of the inherent continuous-time features of CDEs to produce a highly expressive video model. We demonstrate competitive performance against state-of-the-art models for video interpolation and mask propagation tasks.
Authors: Fahim Faisal Niloy, Sk Miraj Ahmed, Dripta S. Raychaudhuri, Samet Oymak, Amit K. Roy-Chowdhury
Traditional test-time adaptation (TTA) methods face significant challenges in adapting to dynamic environments characterized by continuously changing long-term target distributions. These challenges primarily stem from two factors: catastrophic forgetting of previously learned valuable source knowledge and gradual error accumulation caused by miscalibrated pseudo labels. To address these issues, this paper introduces an unsupervised domain change detection method that is capable of identifying domain shifts in dynamic environments and subsequently resets the model parameters to the original source pre-trained values. By restoring the knowledge from the source, it effectively corrects the negative consequences arising from the gradual deterioration of model parameters caused by ongoing shifts in the domain. Our method involves progressive estimation of global batch-norm statistics specific to each domain, while keeping track of changes in the statistics triggered by domain shifts. Importantly, our method is agnostic to the specific adaptation technique employed and thus, can be incorporated to existing TTA methods to enhance their performance in dynamic environments. We perform extensive experiments on benchmark datasets to demonstrate the superior performance of our method compared to state-of-the-art adaptation methods.
Authors: Philip Enevoldsen, Christian Gundersen, Nico Lang, Serge Belongie, Christian Igel
Open-set recognition (OSR), the identification of novel categories, can be a critical component when deploying classification models in real-world applications. Recent work has shown that familiarity-based scoring rules such as the Maximum Softmax Probability (MSP) or the Maximum Logit Score (MLS) are strong baselines when the closed-set accuracy is high. However, one of the potential weaknesses of familiarity-based OSR are adversarial attacks. Here, we present gradient-based adversarial attacks on familiarity scores for both types of attacks, False Familiarity and False Novelty attacks, and evaluate their effectiveness in informed and uninformed settings on TinyImageNet.
Authors: Josué Ruano, Martín Gómez, Eduardo Romero, Antoine Manzanera
Colonoscopy is the choice procedure to diagnose colon and rectum cancer, from early detection of small precancerous lesions (polyps), to confirmation of malign masses. However, the high variability of the organ appearance and the complex shape of both the colon wall and structures of interest make this exploration difficult. Learned visuospatial and perceptual abilities mitigate technical limitations in clinical practice by proper estimation of the intestinal depth. This work introduces a novel methodology to estimate colon depth maps in single frames from monocular colonoscopy videos. The generated depth map is inferred from the shading variation of the colon wall with respect to the light source, as learned from a realistic synthetic database. Briefly, a classic convolutional neural network architecture is trained from scratch to estimate the depth map, improving sharp depth estimations in haustral folds and polyps by a custom loss function that minimizes the estimation error in edges and curvatures. The network was trained by a custom synthetic colonoscopy database herein constructed and released, composed of 248,400 frames (47 videos), with depth annotations at the level of pixels. This collection comprehends 5 subsets of videos with progressively higher levels of visual complexity. Evaluation of the depth estimation with the synthetic database reached a threshold accuracy of 95.65%, and a mean-RMSE of 0.451 cm, while a qualitative assessment with a real database showed consistent depth estimations, visually evaluated by the expert gastroenterologist coauthoring this paper. Finally, the method achieved competitive performance with respect to another state-of-the-art method using a public synthetic database and comparable results in a set of images with other five state-of-the-art methods.
Authors: Robert Johanson, Christian Wilms, Ole Johannsen, Simone Frintrop
Crop detection is integral for precision agriculture applications such as automated yield estimation or fruit picking. However, crop detection, e.g., apple detection in orchard environments remains challenging due to a lack of large-scale datasets and the small relative size of the crops in the image. In this work, we address these challenges by reformulating the apple detection task in a semi-supervised manner. To this end, we provide the large, high-resolution dataset MAD comprising 105 labeled images with 14,667 annotated apple instances and 4,440 unlabeled images. Utilizing this dataset, we also propose a novel Semi-Supervised Small Apple Detection system S$^3$AD based on contextual attention and selective tiling to improve the challenging detection of small apples, while limiting the computational overhead. We conduct an extensive evaluation on MAD and the MSU dataset, showing that S$^3$AD substantially outperforms strong fully-supervised baselines, including several small object detection systems, by up to $14.9\%$. Additionally, we exploit the detailed annotations of our dataset w.r.t. apple properties to analyze the influence of relative size or level of occlusion on the results of various systems, quantifying current challenges.
Authors: Gabriel Efrain Humpire-Mamani, Colin Jacobs, Mathias Prokop, Bram van Ginneken, Nikolas Lessmann
Transfer learning leverages pre-trained model features from a large dataset to save time and resources when training new models for various tasks, potentially enhancing performance. Due to the lack of large datasets in the medical imaging domain, transfer learning from one medical imaging model to other medical imaging models has not been widely explored. This study explores the use of transfer learning to improve the performance of deep convolutional neural networks for organ segmentation in medical imaging. A base segmentation model (3D U-Net) was trained on a large and sparsely annotated dataset; its weights were used for transfer learning on four new down-stream segmentation tasks for which a fully annotated dataset was available. We analyzed the training set size's influence to simulate scarce data. The results showed that transfer learning from the base model was beneficial when small datasets were available, providing significant performance improvements; where fine-tuning the base model is more beneficial than updating all the network weights with vanilla transfer learning. Transfer learning with fine-tuning increased the performance by up to 0.129 (+28\%) Dice score than experiments trained from scratch, and on average 23 experiments increased the performance by 0.029 Dice score in the new segmentation tasks. The study also showed that cross-modality transfer learning using CT scans was beneficial. The findings of this study demonstrate the potential of transfer learning to improve the efficiency of annotation and increase the accessibility of accurate organ segmentation in medical imaging, ultimately leading to improved patient care. We made the network definition and weights publicly available to benefit other users and researchers.
Authors: Hiromu Taketsugu, Norimichi Ukita
Human Pose (HP) estimation is actively researched because of its wide range of applications. However, even estimators pre-trained on large datasets may not perform satisfactorily due to a domain gap between the training and test data. To address this issue, we present our approach combining Active Learning (AL) and Transfer Learning (TL) to adapt HP estimators to individual video domains efficiently. For efficient learning, our approach quantifies (i) the estimation uncertainty based on the temporal changes in the estimated heatmaps and (ii) the unnaturalness in the estimated full-body HPs. These quantified criteria are then effectively combined with the state-of-the-art representativeness criterion to select uncertain and diverse samples for efficient HP estimator learning. Furthermore, we reconsider the existing Active Transfer Learning (ATL) method to introduce novel ideas related to the retraining methods and Stopping Criteria (SC). Experimental results demonstrate that our method enhances learning efficiency and outperforms comparative methods. Our code is publicly available at: https://github.com/ImIntheMiddle/VATL4Pose-WACV2024
Authors: Leonard Salewski, A. Sophia Koepke, Hendrik P. A. Lensch, Zeynep Akata
Converting a model's internals to text can yield human-understandable insights about the model. Inspired by the recent success of training-free approaches for image captioning, we propose ZS-A2T, a zero-shot framework that translates the transformer attention of a given model into natural language without requiring any training. We consider this in the context of Visual Question Answering (VQA). ZS-A2T builds on a pre-trained large language model (LLM), which receives a task prompt, question, and predicted answer, as inputs. The LLM is guided to select tokens which describe the regions in the input image that the VQA model attended to. Crucially, we determine this similarity by exploiting the text-image matching capabilities of the underlying VQA model. Our framework does not require any training and allows the drop-in replacement of different guiding sources (e.g. attribution instead of attention maps), or language models. We evaluate this novel task on textual explanation datasets for VQA, giving state-of-the-art performances for the zero-shot setting on GQA-REX and VQA-X. Our code is available at: https://github.com/ExplainableML/ZS-A2T.
Authors: Daniel Claborne, Eric Slyman, Karl Pazdernik
We train an identity verification architecture and evaluate modifications to the part of the model that combines audio and visual representations, including in scenarios where one input is missing in either of two examples to be compared. We report results on the Voxceleb1-E test set that suggest averaging the output embeddings improves error rate in the full-modality setting and when a single modality is missing, and makes more complete use of the embedding space than systems which use shared layers and discuss possible reasons for this behavior.
Authors: Arindam Dutta, Rohit Lal, Dripta S. Raychaudhuri, Calvin Khang Ta, Amit K. Roy-Chowdhury
Human silhouette extraction is a fundamental task in computer vision with applications in various downstream tasks. However, occlusions pose a significant challenge, leading to incomplete and distorted silhouettes. To address this challenge, we introduce POISE: Pose Guided Human Silhouette Extraction under Occlusions, a novel self-supervised fusion framework that enhances accuracy and robustness in human silhouette prediction. By combining initial silhouette estimates from a segmentation model with human joint predictions from a 2D pose estimation model, POISE leverages the complementary strengths of both approaches, effectively integrating precise body shape information and spatial information to tackle occlusions. Furthermore, the self-supervised nature of \POISE eliminates the need for costly annotations, making it scalable and practical. Extensive experimental results demonstrate its superiority in improving silhouette extraction under occlusions, with promising results in downstream tasks such as gait recognition. The code for our method is available https://github.com/take2rohit/poise.
Authors: Xinyu Zhang, Weiyu Sun, Hao Lu, Ying Chen, Yun Ge, Xiaolin Huang, Jie Yuan, Yingcong Chen
Remote photoplethysmography (rPPG) is a noninvasive technique that aims to capture subtle variations in facial pixels caused by changes in blood volume resulting from cardiac activities. Most existing unsupervised methods for rPPG tasks focus on the contrastive learning between samples while neglecting the inherent self-similar prior in physiological signals. In this paper, we propose a Self-Similarity Prior Distillation (SSPD) framework for unsupervised rPPG estimation, which capitalizes on the intrinsic self-similarity of cardiac activities. Specifically, we first introduce a physical-prior embedded augmentation technique to mitigate the effect of various types of noise. Then, we tailor a self-similarity-aware network to extract more reliable self-similar physiological features. Finally, we develop a hierarchical self-distillation paradigm to assist the network in disentangling self-similar physiological patterns from facial videos. Comprehensive experiments demonstrate that the unsupervised SSPD framework achieves comparable or even superior performance compared to the state-of-the-art supervised methods. Meanwhile, SSPD maintains the lowest inference time and computation cost among end-to-end models. The source codes are available at https://github.com/LinXi1C/SSPD.
Authors: Kartik Gupta, Akshay Asthana
Quantized networks use less computational and memory resources and are suitable for deployment on edge devices. While quantization-aware training QAT is the well-studied approach to quantize the networks at low precision, most research focuses on over-parameterized networks for classification with limited studies on popular and edge device friendly single-shot object detection and semantic segmentation methods like YOLO. Moreover, majority of QAT methods rely on Straight-through Estimator (STE) approximation which suffers from an oscillation phenomenon resulting in sub-optimal network quantization. In this paper, we show that it is difficult to achieve extremely low precision (4-bit and lower) for efficient YOLO models even with SOTA QAT methods due to oscillation issue and existing methods to overcome this problem are not effective on these models. To mitigate the effect of oscillation, we first propose Exponentially Moving Average (EMA) based update to the QAT model. Further, we propose a simple QAT correction method, namely QC, that takes only a single epoch of training after standard QAT procedure to correct the error induced by oscillating weights and activations resulting in a more accurate quantized model. With extensive evaluation on COCO dataset using various YOLO5 and YOLO7 variants, we show that our correction method improves quantized YOLO networks consistently on both object detection and segmentation tasks at low-precision (4-bit and 3-bit).
Authors: Zixun Zhang, Yuncheng Jiang, Jun Wei, Hannah Cui, Zhen Li
Automatic polyp segmentation models play a pivotal role in the clinical diagnosis of gastrointestinal diseases. In previous studies, most methods relied on fully supervised approaches, necessitating pixel-level annotations for model training. However, the creation of pixel-level annotations is both expensive and time-consuming, impeding the development of model generalization. In response to this challenge, we introduce ScribblePolyp, a novel scribble-supervised polyp segmentation framework. Unlike fully-supervised models, ScribblePolyp only requires the annotation of two lines (scribble labels) for each image, significantly reducing the labeling cost. Despite the coarse nature of scribble labels, which leave a substantial portion of pixels unlabeled, we propose a two-branch consistency alignment approach to provide supervision for these unlabeled pixels. The first branch employs transformation consistency alignment to narrow the gap between predictions under different transformations of the same input image. The second branch leverages affinity propagation to refine predictions into a soft version, extending additional supervision to unlabeled pixels. In summary, ScribblePolyp is an efficient model that does not rely on teacher models or moving average pseudo labels during training. Extensive experiments on the SUN-SEG dataset underscore the effectiveness of ScribblePolyp, achieving a Dice score of 0.8155, with the potential for a 1.8% improvement in the Dice score through a straightforward self-training strategy.
Authors: Rui Xu, Wenkang Qin, Peixiang Huang, Haowang, Lin Luo
Deep Neural Networks (DNNs) are expected to provide explanation for users to understand their black-box predictions. Saliency map is a common form of explanation illustrating the heatmap of feature attributions, but it suffers from noise in distinguishing important features. In this paper, we propose a model-agnostic learning method called Saliency Constrained Adaptive Adversarial Training (SCAAT) to improve the quality of such DNN interpretability. By constructing adversarial samples under the guidance of saliency map, SCAAT effectively eliminates most noise and makes saliency maps sparser and more faithful without any modification to the model architecture. We apply SCAAT to multiple DNNs and evaluate the quality of the generated saliency maps on various natural and pathological image datasets. Evaluations on different domains and metrics show that SCAAT significantly improves the interpretability of DNNs by providing more faithful saliency maps without sacrificing their predictive power.
Authors: Rishav Bhardwaj, Janarthanam Jothi Balaji, Vasudevan Lakshminarayanan
There has been considerable progress in implicit neural representation to upscale an image to any arbitrary resolution. However, existing methods are based on defining a function to predict the Red, Green and Blue (RGB) value from just four specific loci. Relying on just four loci is insufficient as it leads to losing fine details from the neighboring region(s). We show that by taking into account the semi-local region leads to an improvement in performance. In this paper, we propose applying a new technique called Overlapping Windows on Semi-Local Region (OW-SLR) to an image to obtain any arbitrary resolution by taking the coordinates of the semi-local region around a point in the latent space. This extracted detail is used to predict the RGB value of a point. We illustrate the technique by applying the algorithm to the Optical Coherence Tomography-Angiography (OCT-A) images and show that it can upscale them to random resolution. This technique outperforms the existing state-of-the-art methods when applied to the OCT500 dataset. OW-SLR provides better results for classifying healthy and diseased retinal images such as diabetic retinopathy and normals from the given set of OCT-A images. The project page is available at https://rishavbb.github.io/ow-slr/index.html
Authors: Kui Jiang, Xuemei Jia, Wenxin Huang, Wenbin Wang, Zheng Wang, Junjun Jiang
CNNs and Self attention have achieved great success in multimedia applications for dynamic association learning of self-attention and convolution in image restoration. However, CNNs have at least two shortcomings: 1) limited receptive field; 2) static weight of sliding window at inference, unable to cope with the content diversity.In view of the advantages and disadvantages of CNNs and Self attention, this paper proposes an association learning method to utilize the advantages and suppress their shortcomings, so as to achieve high-quality and efficient inpainting. We regard rain distribution reflects the degradation location and degree, in addition to the rain distribution prediction. Thus, we propose to refine background textures with the predicted degradation prior in an association learning manner. As a result, we accomplish image deraining by associating rain streak removal and background recovery, where an image deraining network and a background recovery network are designed for two subtasks. The key part of association learning is a novel multi-input attention module. It generates the degradation prior and produces the degradation mask according to the predicted rainy distribution. Benefited from the global correlation calculation of SA, MAM can extract the informative complementary components from the rainy input with the degradation mask, and then help accurate texture restoration. Meanwhile, SA tends to aggregate feature maps with self-attention importance, but convolution diversifies them to focus on the local textures. A hybrid fusion network involves one residual Transformer branch and one encoder-decoder branch. The former takes a few learnable tokens as input and stacks multi-head attention and feed-forward networks to encode global features of the image. The latter, conversely, leverages the multi-scale encoder-decoder to represent contexture knowledge.
Authors: Haoyi Duan, Yan Xia, Mingze Zhou, Li Tang, Jieming Zhu, Zhou Zhao
In recent years, the deployment of large-scale pre-trained models in audio-visual downstream tasks has yielded remarkable outcomes. However, these models, primarily trained on single-modality unconstrained datasets, still encounter challenges in feature extraction for multi-modal tasks, leading to suboptimal performance. This limitation arises due to the introduction of irrelevant modality-specific information during encoding, which adversely affects the performance of downstream tasks. To address this challenge, this paper proposes a novel Dual-Guided Spatial-Channel-Temporal (DG-SCT) attention mechanism. This mechanism leverages audio and visual modalities as soft prompts to dynamically adjust the parameters of pre-trained models based on the current multi-modal input features. Specifically, the DG-SCT module incorporates trainable cross-modal interaction layers into pre-trained audio-visual encoders, allowing adaptive extraction of crucial information from the current modality across spatial, channel, and temporal dimensions, while preserving the frozen parameters of large-scale pre-trained models. Experimental evaluations demonstrate that our proposed model achieves state-of-the-art results across multiple downstream tasks, including AVE, AVVP, AVS, and AVQA. Furthermore, our model exhibits promising performance in challenging few-shot and zero-shot scenarios. The source code and pre-trained models are available at https://github.com/haoyi-duan/DG-SCT.
Authors: Qinghua Lin, Zuoyong Li, Kun Zeng, Haoyi Fan, Wei Li, Xiaoguang Zhou
Deep learning techniques have greatly enhanced the performance of fire detection in videos. However, video-based fire detection models heavily rely on labeled data, and the process of data labeling is particularly costly and time-consuming, especially when dealing with videos. Considering the limited quantity of labeled video data, we propose a semi-supervised fire detection model called FireMatch, which is based on consistency regularization and adversarial distribution alignment. Specifically, we first combine consistency regularization with pseudo-label. For unlabeled data, we design video data augmentation to obtain corresponding weakly augmented and strongly augmented samples. The proposed model predicts weakly augmented samples and retains pseudo-label above a threshold, while training on strongly augmented samples to predict these pseudo-labels for learning more robust feature representations. Secondly, we generate video cross-set augmented samples by adversarial distribution alignment to expand the training data and alleviate the decline in classification performance caused by insufficient labeled data. Finally, we introduce a fairness loss to help the model produce diverse predictions for input samples, thereby addressing the issue of high confidence with the non-fire class in fire classification scenarios. The FireMatch achieved an accuracy of 76.92% and 91.81% on two real-world fire datasets, respectively. The experimental results demonstrate that the proposed method outperforms the current state-of-the-art semi-supervised classification methods.
Authors: Yuxin Zhu, Xilei Zhu, Huiyu Duan, Jie Li, Kaiwei Zhang, Yucheng Zhu, Li Chen, Xiongkuo Min, Guangtao Zhai
Visual saliency prediction for omnidirectional videos (ODVs) has shown great significance and necessity for omnidirectional videos to help ODV coding, ODV transmission, ODV rendering, etc.. However, most studies only consider visual information for ODV saliency prediction while audio is rarely considered despite its significant influence on the viewing behavior of ODV. This is mainly due to the lack of large-scale audio-visual ODV datasets and corresponding analysis. Thus, in this paper, we first establish the largest audio-visual saliency dataset for omnidirectional videos (AVS-ODV), which comprises the omnidirectional videos, audios, and corresponding captured eye-tracking data for three video sound modalities including mute, mono, and ambisonics. Then we analyze the visual attention behavior of the observers under various omnidirectional audio modalities and visual scenes based on the AVS-ODV dataset. Furthermore, we compare the performance of several state-of-the-art saliency prediction models on the AVS-ODV dataset and construct a new benchmark. Our AVS-ODV datasets and the benchmark will be released to facilitate future research.
Authors: Hoang C. Nguyen, Chi Phan, Hieu H. Pham
Screening mammography is the most widely used method for early breast cancer detection, significantly reducing mortality rates. The integration of information from multi-view mammograms enhances radiologists' confidence and diminishes false-positive rates since they can examine on dual-view of the same breast to cross-reference the existence and location of the lesion. Inspired by this, we present TransReg, a Computer-Aided Detection (CAD) system designed to exploit the relationship between craniocaudal (CC), and mediolateral oblique (MLO) views. The system includes cross-transformer to model the relationship between the region of interest (RoIs) extracted by siamese Faster RCNN network for mass detection problems. Our work is the first time cross-transformer has been integrated into an object detection framework to model the relation between ipsilateral views. Our experimental evaluation on DDSM and VinDr-Mammo datasets shows that our TransReg, equipped with SwinT as a feature extractor achieves state-of-the-art performance. Specifically, at the false positive rate per image at 0.5, TransReg using SwinT gets a recall at 83.3% for DDSM dataset and 79.7% for VinDr-Mammo dataset. Furthermore, we conduct a comprehensive analysis to demonstrate that cross-transformer can function as an auto-registration module, aligning the masses in dual-view and utilizing this information to inform final predictions. It is a replication diagnostic workflow of expert radiologists
Authors: Fabiha Bushra, Muhammad E. H. Chowdhury, Rusab Sarmun, Saidul Kabir, Menatalla Said, Sohaib Bassam Zoghoul, Adam Mushtak, Israa Al-Hashimi, Abdulrahman Alqahtani, Anwarul Hasan
Pulmonary Embolism (PE) is a critical medical condition characterized by obstructions in the pulmonary arteries. Despite being a major health concern, it often goes underdiagnosed leading to detrimental clinical outcomes. The increasing reliance on Computed Tomography Pulmonary Angiography for diagnosis presents challenges and a pressing need for enhanced diagnostic solutions. The primary objective of this study is to leverage deep learning techniques to enhance the Computer Assisted Diagnosis of PE. This study presents a comprehensive dual-pronged approach combining classification and detection for PE diagnosis. We introduce an Attention-Guided Convolutional Neural Network (AG-CNN) for classification, addressing both global and local lesion region. For detection, state-of-the-art models are employed to pinpoint potential PE regions. Different ensembling techniques further improve detection accuracy by combining predictions from different models. Finally, a heuristic strategy integrates classifier outputs with detection results, ensuring robust and accurate PE identification. Our attention-guided classification approach, tested on the Ferdowsi University of Mashhad's Pulmonary Embolism (FUMPE) dataset, outperformed the baseline model DenseNet-121 by achieving an 8.1% increase in the Area Under the Receiver Operating Characteristic. By employing ensemble techniques with detection models, the mean average precision (mAP) was considerably enhanced by a 4.7% increase. The classifier-guided framework further refined the mAP and F1 scores over the ensemble models. Our research offers a comprehensive approach to PE diagnostics using deep learning, addressing the prevalent issues of underdiagnosis and misdiagnosis. We aim to improve PE patient care by integrating AI solutions into clinical workflows, highlighting the potential of human-AI collaboration in medical diagnostics.
Authors: Jay Gala, Sauradip Nag, Huichou Huang, Ruirui Liu, Xiatian Zhu
Cloud analysis is a critical component of weather and climate science, impacting various sectors like disaster management. However, achieving fine-grained cloud analysis, such as cloud segmentation, in remote sensing remains challenging due to the inherent difficulties in obtaining accurate labels, leading to significant labeling errors in training data. Existing methods often assume the availability of reliable segmentation annotations, limiting their overall performance. To address this inherent limitation, we introduce an innovative model-agnostic Cloud Adaptive-Labeling (CAL) approach, which operates iteratively to enhance the quality of training data annotations and consequently improve the performance of the learned model. Our methodology commences by training a cloud segmentation model using the original annotations. Subsequently, it introduces a trainable pixel intensity threshold for adaptively labeling the cloud training images on the fly. The newly generated labels are then employed to fine-tune the model. Extensive experiments conducted on multiple standard cloud segmentation benchmarks demonstrate the effectiveness of our approach in significantly boosting the performance of existing segmentation models. Our CAL method establishes new state-of-the-art results when compared to a wide array of existing alternatives.
Authors: Yongcheng Zong, Shuqiang Wang
Brain network analysis has emerged as pivotal method for gaining a deeper understanding of brain functions and disease mechanisms. Despite the existence of various network construction approaches, shortcomings persist in the learning of correlations between structural and functional brain imaging data. In light of this, we introduce a novel method called BrainNetDiff, which combines a multi-head Transformer encoder to extract relevant features from fMRI time series and integrates a conditional latent diffusion model for brain network generation. Leveraging a conditional prompt and a fusion attention mechanism, this method significantly improves the accuracy and stability of brain network generation. To the best of our knowledge, this represents the first framework that employs diffusion for the fusion of the multimodal brain imaging and brain network generation from images to graphs. We validate applicability of this framework in the construction of brain network across healthy and neurologically impaired cohorts using the authentic dataset. Experimental results vividly demonstrate the significant effectiveness of the proposed method across the downstream disease classification tasks. These findings convincingly emphasize the prospective value in the field of brain network research, particularly its key significance in neuroimaging analysis and disease diagnosis. This research provides a valuable reference for the processing of multimodal brain imaging data and introduces a novel, efficient solution to the field of neuroimaging.
Authors: Tim Büchner, Sven Sickert, Gerd Fabian Volk, Christoph Anders, Orlando Guntinas-Lichius, Joachim Denzler
The human face is one of the most crucial parts in interhuman communication. Even when parts of the face are hidden or obstructed the underlying facial movements can be understood. Machine learning approaches often fail in that regard due to the complexity of the facial structures. To alleviate this problem a common approach is to fine-tune a model for such a specific application. However, this is computational intensive and might have to be repeated for each desired analysis task. In this paper, we propose to reconstruct obstructed facial parts to avoid the task of repeated fine-tuning. As a result, existing facial analysis methods can be used without further changes with respect to the data. In our approach, the restoration of facial features is interpreted as a style transfer task between different recording setups. By using the CycleGAN architecture the requirement of matched pairs, which is often hard to fullfill, can be eliminated. To proof the viability of our approach, we compare our reconstructions with real unobstructed recordings. We created a novel data set in which 36 test subjects were recorded both with and without 62 surface electromyography sensors attached to their faces. In our evaluation, we feature typical facial analysis tasks, like the computation of Facial Action Units and the detection of emotions. To further assess the quality of the restoration, we also compare perceptional distances. We can show, that scores similar to the videos without obstructing sensors can be achieved.
Authors: Senthil Purushwalkam, Nikhil Naik
We present a novel method for reconstructing 3D objects from a single RGB image. Our method leverages the latest image generation models to infer the hidden 3D structure while remaining faithful to the input image. While existing methods obtain impressive results in generating 3D models from text prompts, they do not provide an easy approach for conditioning on input RGB data. Na\"ive extensions of these methods often lead to improper alignment in appearance between the input image and the 3D reconstructions. We address these challenges by introducing Image Constrained Radiance Fields (ConRad), a novel variant of neural radiance fields. ConRad is an efficient 3D representation that explicitly captures the appearance of an input image in one viewpoint. We propose a training algorithm that leverages the single RGB image in conjunction with pretrained Diffusion Models to optimize the parameters of a ConRad representation. Extensive experiments show that ConRad representations can simplify preservation of image details while producing a realistic 3D reconstruction. Compared to existing state-of-the-art baselines, we show that our 3D reconstructions remain more faithful to the input and produce more consistent 3D models while demonstrating significantly improved quantitative performance on a ShapeNet object benchmark.
Authors: Shuhei Tarashima, Muhammad Abdul Haq, Yushan Wang, Norio Tagawa
In this work, we present a novel Sports Ball Detection and Tracking (SBDT) method that can be applied to various sports categories. Our approach is composed of (1) high-resolution feature extraction, (2) position-aware model training, and (3) inference considering temporal consistency, all of which are put together as a new SBDT baseline. Besides, to validate the wide-applicability of our approach, we compare our baseline with 6 state-of-the-art SBDT methods on 5 datasets from different sports categories. We achieve this by newly introducing two SBDT datasets, providing new ball annotations for two datasets, and re-implementing all the methods to ease extensive comparison. Experimental results demonstrate that our approach is substantially superior to existing methods on all the sports categories covered by the datasets. We believe our proposed method can play as a Widely Applicable Strong Baseline (WASB) of SBDT, and our datasets and codebase will promote future SBDT research. Datasets and codes will be made publicly available.
Authors: Ajinkya Kadu, Felix Lucka, Kees Joost Batenburg
This paper presents a novel method for the reconstruction of high-resolution temporal images in dynamic tomographic imaging, particularly for discrete objects with smooth boundaries that vary over time. Addressing the challenge of limited measurements per time point, we propose a technique that synergistically incorporates spatial and temporal information of the dynamic objects. This is achieved through the application of the level-set method for image segmentation and the representation of motion via a sinusoidal basis. The result is a computationally efficient and easily optimizable variational framework that enables the reconstruction of high-quality 2D or 3D image sequences with a single projection per frame. Compared to current methods, our proposed approach demonstrates superior performance on both synthetic and pseudo-dynamic real X-ray tomography datasets. The implications of this research extend to improved visualization and analysis of dynamic processes in tomographic imaging, finding potential applications in diverse scientific and industrial domains.
Authors: Haokun Zhu, Juang Ian Chong, Teng Hu, Ran Yi, Yu-Kun Lai, Paul L. Rosin
Vector graphics are widely used in graphical designs and have received more and more attention. However, unlike raster images which can be easily obtained, acquiring high-quality vector graphics, typically through automatically converting from raster images remains a significant challenge, especially for more complex images such as photos or artworks. In this paper, we propose SAMVG, a multi-stage model to vectorize raster images into SVG (Scalable Vector Graphics). Firstly, SAMVG uses general image segmentation provided by the Segment-Anything Model and uses a novel filtering method to identify the best dense segmentation map for the entire image. Secondly, SAMVG then identifies missing components and adds more detailed components to the SVG. Through a series of extensive experiments, we demonstrate that SAMVG can produce high quality SVGs in any domain while requiring less computation time and complexity compared to previous state-of-the-art methods.
Authors: Sen Wang, Wei Zhang, Stefano Gasperini, Shun-Cheng Wu, Nassir Navab
Creating high-quality view synthesis is essential for immersive applications but continues to be problematic, particularly in indoor environments and for real-time deployment. Current techniques frequently require extensive computational time for both training and rendering, and often produce less-than-ideal 3D representations due to inadequate geometric structuring. To overcome this, we introduce VoxNeRF, a novel approach that leverages volumetric representations to enhance the quality and efficiency of indoor view synthesis. Firstly, VoxNeRF constructs a structured scene geometry and converts it into a voxel-based representation. We employ multi-resolution hash grids to adaptively capture spatial features, effectively managing occlusions and the intricate geometry of indoor scenes. Secondly, we propose a unique voxel-guided efficient sampling technique. This innovation selectively focuses computational resources on the most relevant portions of ray segments, substantially reducing optimization time. We validate our approach against three public indoor datasets and demonstrate that VoxNeRF outperforms state-of-the-art methods. Remarkably, it achieves these gains while reducing both training and rendering times, surpassing even Instant-NGP in speed and bringing the technology closer to real-time.
Authors: Cheng Yang, Rui Xu, Ye Guo, Peixiang Huang, Yiru Chen, Wenkui Ding, Zhongyuan Wang, Hong Zhou
Visual commonsense reasoning (VCR) is a challenging multi-modal task, which requires high-level cognition and commonsense reasoning ability about the real world. In recent years, large-scale pre-training approaches have been developed and promoted the state-of-the-art performance of VCR. However, the existing approaches almost employ the BERT-like objectives to learn multi-modal representations. These objectives motivated from the text-domain are insufficient for the excavation on the complex scenario of visual modality. Most importantly, the spatial distribution of the visual objects is basically neglected. To address the above issue, we propose to construct the spatial relation graph based on the given visual scenario. Further, we design two pre-training tasks named object position regression (OPR) and spatial relation classification (SRC) to learn to reconstruct the spatial relation graph respectively. Quantitative analysis suggests that the proposed method can guide the representations to maintain more spatial context and facilitate the attention on the essential visual regions for reasoning. We achieve the state-of-the-art results on VCR and two other vision-and-language reasoning tasks VQA, and NLVR.
Authors: Arunkumar Rathinam, Haytam Qadadri, Djamila Aouada
In recent years, there has been a growing demand for improved autonomy for in-orbit operations such as rendezvous, docking, and proximity maneuvers, leading to increased interest in employing Deep Learning-based Spacecraft Pose Estimation techniques. However, due to limited access to real target datasets, algorithms are often trained using synthetic data and applied in the real domain, resulting in a performance drop due to the domain gap. State-of-the-art approaches employ Domain Adaptation techniques to mitigate this issue. In the search for viable solutions, event sensing has been explored in the past and shown to reduce the domain gap between simulations and real-world scenarios. Event sensors have made significant advancements in hardware and software in recent years. Moreover, the characteristics of the event sensor offer several advantages in space applications compared to RGB sensors. To facilitate further training and evaluation of DL-based models, we introduce a novel dataset, SPADES, comprising real event data acquired in a controlled laboratory environment and simulated event data using the same camera intrinsics. Furthermore, we propose an effective data filtering method to improve the quality of training data, thus enhancing model performance. Additionally, we introduce an image-based event representation that outperforms existing representations. A multifaceted baseline evaluation was conducted using different event representations, event filtering strategies, and algorithmic frameworks, and the results are summarized. The dataset will be made available at this http URL
Authors: Sihan Gao, Jing Zhu, Xiaoxuan Zhuang, Zhaoyue Wang, Qijin Li
In recent years, human pose estimation has made significant progress through the implementation of deep learning techniques. However, these techniques still face limitations when confronted with challenging scenarios, including occlusion, diverse appearances, variations in illumination, and overlap. To cope with such drawbacks, we present the Spatial Attention-based Distribution Integration Network (SADI-NET) to improve the accuracy of localization in such situations. Our network consists of three efficient models: the receptive fortified module (RFM), spatial fusion module (SFM), and distribution learning module (DLM). Building upon the classic HourglassNet architecture, we replace the basic block with our proposed RFM. The RFM incorporates a dilated residual block and attention mechanism to expand receptive fields while enhancing sensitivity to spatial information. In addition, the SFM incorporates multi-scale characteristics by employing both global and local attention mechanisms. Furthermore, the DLM, inspired by residual log-likelihood estimation (RLE), reconfigures a predicted heatmap using a trainable distribution weight. For the purpose of determining the efficacy of our model, we conducted extensive experiments on the MPII and LSP benchmarks. Particularly, our model obtained a remarkable $92.10\%$ percent accuracy on the MPII test dataset, demonstrating significant improvements over existing models and establishing state-of-the-art performance.
Authors: Licheng Wen, Xuemeng Yang, Daocheng Fu, Xiaofeng Wang, Pinlong Cai, Xin Li, Tao Ma, Yingxuan Li, Linran Xu, Dengke Shang, Zheng Zhu, Shaoyan Sun, Yeqi Bai, Xinyu Cai, Min Dou, Shuanglu Hu, Botian Shi
The pursuit of autonomous driving technology hinges on the sophisticated integration of perception, decision-making, and control systems. Traditional approaches, both data-driven and rule-based, have been hindered by their inability to grasp the nuance of complex driving environments and the intentions of other road users. This has been a significant bottleneck, particularly in the development of common sense reasoning and nuanced scene understanding necessary for safe and reliable autonomous driving. The advent of Visual Language Models (VLM) represents a novel frontier in realizing fully autonomous vehicle driving. This report provides an exhaustive evaluation of the latest state-of-the-art VLM, \modelnamefull, and its application in autonomous driving scenarios. We explore the model's abilities to understand and reason about driving scenes, make decisions, and ultimately act in the capacity of a driver. Our comprehensive tests span from basic scene recognition to complex causal reasoning and real-time decision-making under varying conditions. Our findings reveal that \modelname demonstrates superior performance in scene understanding and causal reasoning compared to existing autonomous systems. It showcases the potential to handle out-of-distribution scenarios, recognize intentions, and make informed decisions in real driving contexts. However, challenges remain, particularly in direction discernment, traffic light recognition, vision grounding, and spatial reasoning tasks. These limitations underscore the need for further research and development. Project is now available on GitHub for interested parties to access and utilize: \url{https://github.com/PJLab-ADG/GPT4V-AD-Exploration}
Authors: Meiling Fang, Marco Huber, Julian Fierrez, Raghavendra Ramachandra, Naser Damer, Alhasan Alkhaddour, Maksim Kasantcev, Vasiliy Pryadchenko, Ziyuan Yang, Huijie Huangfu, Yingyu Chen, Yi Zhang, Yuchen Pan, Junjun Jiang, Xianming Liu, Xianyun Sun, Caiyong Wang, Xingyu Liu, Zhaohua Chang, Guangzhe Zhao, Juan Tapia, Lazaro Gonzalez-Soler, Carlos Aravena, Daniel Schulz
This paper presents a summary of the Competition on Face Presentation Attack Detection Based on Privacy-aware Synthetic Training Data (SynFacePAD 2023) held at the 2023 International Joint Conference on Biometrics (IJCB 2023). The competition attracted a total of 8 participating teams with valid submissions from academia and industry. The competition aimed to motivate and attract solutions that target detecting face presentation attacks while considering synthetic-based training data motivated by privacy, legal and ethical concerns associated with personal data. To achieve that, the training data used by the participants was limited to synthetic data provided by the organizers. The submitted solutions presented innovations and novel approaches that led to outperforming the considered baseline in the investigated benchmarks.
Authors: Jinjin Xu, Liwu Xu, Yuzhe Yang, Xiang Li, Yanchun Xie, Yi-Jie Huang, Yaqian Li
Recent advances such as LLaVA and Mini-GPT4 have successfully integrated visual information into LLMs, yielding inspiring outcomes and giving rise to a new generation of multi-modal LLMs, or MLLMs. Nevertheless, these methods struggle with hallucinations and the mutual interference between tasks. To tackle these problems, we propose an efficient and accurate approach to adapt to downstream tasks by utilizing LLM as a bridge to connect multiple expert models, namely u-LLaVA. Firstly, we incorporate the modality alignment module and multi-task modules into LLM. Then, we reorganize or rebuild multi-type public datasets to enable efficient modality alignment and instruction following. Finally, task-specific information is extracted from the trained LLM and provided to different modules for solving downstream tasks. The overall framework is simple, effective, and achieves state-of-the-art performance across multiple benchmarks. We also release our model, the generated data, and the code base publicly available.
Authors: Yuxuan Ou, Yuzhe Zhang, Yuntang Wang, Shwetak Patel, Daniel McDuf, Xin Liu
Recent advances in supervised deep learning techniques have demonstrated the possibility to remotely measure human physiological vital signs (e.g., photoplethysmograph, heart rate) just from facial videos. However, the performance of these methods heavily relies on the availability and diversity of real labeled data. Yet, collecting large-scale real-world data with high-quality labels is typically challenging and resource intensive, which also raises privacy concerns when storing personal bio-metric data. Synthetic video-based datasets (e.g., SCAMPS~\cite{mcduff2022scamps}) with photo-realistic synthesized avatars are introduced to alleviate the issues while providing high-quality synthetic data. However, there exists a significant gap between synthetic and real-world data, which hinders the generalization of neural models trained on these synthetic datasets. In this paper, we proposed several measures to add real-world noise to synthetic physiological signals and corresponding facial videos. We experimented with individual and combined augmentation methods and evaluated our framework on three public real-world datasets. Our results show that we were able to reduce the average MAE from 6.9 to 2.0.
Authors: Wojciech Michal Matkowski, Xiaojie Li, Adams Wai Kin Kong
The prevalence of smartphone and consumer camera has led to more evidence in the form of digital images, which are mostly taken in uncontrolled and uncooperative environments. In these images, criminals likely hide or cover their faces while their hands are observable in some cases, creating a challenging use case for forensic investigation. Many existing hand-based recognition methods perform well for hand images collected in controlled environments with user cooperation. However, their performance deteriorates significantly in uncontrolled and uncooperative environments. A recent work has exposed the potential of hand recognition in these environments. However, only the palmar regions were considered, and the recognition performance is still far from satisfactory. To improve the recognition accuracy, an algorithm integrating a multi-spatial transformer network (MSTN) and multiple loss functions is proposed to fully utilize information in full hand images. MSTN is firstly employed to localize the palms and fingers and estimate the alignment parameters. Then, the aligned images are further fed into pretrained convolutional neural networks, where features are extracted. Finally, a training scheme with multiple loss functions is used to train the network end-to-end. To demonstrate the effectiveness of the proposed algorithm, the trained model is evaluated on NTU-PI-v1 database and six benchmark databases from different domains. Experimental results show that the proposed algorithm performs significantly better than the existing methods in these uncontrolled and uncooperative environments and has good generalization capabilities to samples from different domains.
Authors: Dieuwertje Alblas, Julian Suk, Christoph Brune, Kak Khee Yeung, Jelmer M. Wolterink
Blood vessel orientation as visualized in 3D medical images is an important descriptor of its geometry that can be used for centerline extraction and subsequent segmentation and visualization. Arteries appear at many scales and levels of tortuosity, and determining their exact orientation is challenging. Recent works have used 3D convolutional neural networks (CNNs) for this purpose, but CNNs are sensitive to varying vessel sizes and orientations. We present SIRE: a scale-invariant, rotation-equivariant estimator for local vessel orientation. SIRE is modular and can generalise due to symmetry preservation.
SIRE consists of a gauge equivariant mesh CNN (GEM-CNN) operating on multiple nested spherical meshes with different sizes in parallel. The features on each mesh are a projection of image intensities within the corresponding sphere. These features are intrinsic to the sphere and, in combination with the GEM-CNN, lead to SO(3)-equivariance. Approximate scale invariance is achieved by weight sharing and use of a symmetric maximum function to combine multi-scale predictions. Hence, SIRE can be trained with arbitrarily oriented vessels with varying radii to generalise to vessels with a wide range of calibres and tortuosity.
We demonstrate the efficacy of SIRE using three datasets containing vessels of varying scales: the vascular model repository (VMR), the ASOCA coronary artery set, and a set of abdominal aortic aneurysms (AAAs). We embed SIRE in a centerline tracker which accurately tracks AAAs, regardless of the data SIRE is trained with. Moreover, SIRE can be used to track coronary arteries, even when trained only with AAAs.
In conclusion, by incorporating SO(3) and scale symmetries, SIRE can determine the orientations of vessels outside of the training domain, forming a robust and data-efficient solution to geometric analysis of blood vessels in 3D medical images.
Authors: Zhen Zhou, Yunkai Ma, Junfeng Fan, Zhaoyang Liu, Fengshui Jing, Min Tan
Due to the frequent variability of object orientation, accurate prediction of orientation information remains a challenge in oriented object detection. To better extract orientation-related information, current methods primarily focus on the design of reasonable representations of oriented bounding box (OBB) and rotation-sensitive feature extraction. However, existing OBB representations often suffer from boundary discontinuity and representation ambiguity problems. Methods of designing continuous and unambiguous regression losses do not essentially solve such problems. Gaussian bounding box (GBB) avoids these OBB representation problems, but directly regressing GBB is susceptible to numerical instability. In this paper, we propose linear GBB (LGBB), a novel OBB representation. By linearly transforming the elements of GBB, LGBB does not have the boundary discontinuity and representation ambiguity problems, and have high numerical stability. On the other hand, current rotation-sensitive feature extraction methods based on convolutions can only extract features under a local receptive field, which is slow in aggregating rotation-sensitive features. To address this issue, we propose ring-shaped rotated convolution (RRC). By adaptively rotating feature maps to arbitrary orientations, RRC extracts rotation-sensitive features under a ring-shaped receptive field, rapidly aggregating rotation-sensitive features and contextual information. RRC can be applied to various models in a plug-and-play manner. Experimental results demonstrate that the proposed LGBB and RRC are effective and achieve state-of-the-art (SOTA) performance. By integrating LGBB and RRC into various models, the detection accuracy is effectively improved on DOTA and HRSC2016 datasets.
Authors: Yongfeng Chena, Jin Liua, Zhijing Yang, Ruihan Chena, Junpeng Tan
Recently, commonsense learning has been a hot topic in image-text matching. Although it can describe more graphic correlations, commonsense learning still has some shortcomings: 1) The existing methods are based on triplet semantic similarity measurement loss, which cannot effectively match the intractable negative in image-text sample pairs. 2) The weak generalization ability of the model leads to the poor effect of image and text matching on large-scale datasets. According to these shortcomings. This paper proposes a novel image-text matching model, called Active Mining Sample Pair Semantics image-text matching model (AMSPS). Compared with the single semantic learning mode of the commonsense learning model with triplet loss function, AMSPS is an active learning idea. Firstly, the proposed Adaptive Hierarchical Reinforcement Loss (AHRL) has diversified learning modes. Its active learning mode enables the model to more focus on the intractable negative samples to enhance the discriminating ability. In addition, AMSPS can also adaptively mine more hidden relevant semantic representations from uncommented items, which greatly improves the performance and generalization ability of the model. Experimental results on Flickr30K and MSCOCO universal datasets show that our proposed method is superior to advanced comparison methods.
Authors: ShiQi Jiang, JunJie Kang, YuJian Li
The color and texture dual pipeline architecture (CTDP) suppresses texture representation and artifacts through masked total variation loss (Mtv), and further experiments have shown that smooth input can almost completely eliminate texture representation. We have demonstrated through experiments that smooth input is not the key reason for removing texture representations, but rather the distribution differentiation of the training dataset. Based on this, we propose an input distribution differentiation training strategy (IDD), which forces the generation of textures to be completely dependent on the noise distribution, while the smooth distribution will not produce textures at all. Overall, our proposed distribution differentiation training strategy allows for two pre-defined input distributions to be responsible for two generation tasks, with noise distribution responsible for texture generation and smooth distribution responsible for color smooth transfer. Finally, we choose a smooth distribution as the input for the forward inference stage to completely eliminate texture representations and artifacts in color transfer tasks.
Authors: Shilong Liu, Hao Cheng, Haotian Liu, Hao Zhang, Feng Li, Tianhe Ren, Xueyan Zou, Jianwei Yang, Hang Su, Jun Zhu, Lei Zhang, Jianfeng Gao, Chunyuan Li
LLaVA-Plus is a general-purpose multimodal assistant that expands the capabilities of large multimodal models. It maintains a skill repository of pre-trained vision and vision-language models and can activate relevant tools based on users' inputs to fulfill real-world tasks. LLaVA-Plus is trained on multimodal instruction-following data to acquire the ability to use tools, covering visual understanding, generation, external knowledge retrieval, and compositions. Empirical results show that LLaVA-Plus outperforms LLaVA in existing capabilities and exhibits new ones. It is distinct in that the image query is directly grounded and actively engaged throughout the entire human-AI interaction sessions, significantly improving tool use performance and enabling new scenarios.
Authors: Adam J Shephard, Hanya Mahmood, Shan E Ahmed Raza, Anna Luiza Damaceno Araujo, Alan Roger Santos-Silva, Marcio Ajudarte Lopes, Pablo Agustin Vargas, Kris McCombe, Stephanie Craig, Jacqueline James, Jill Brooks, Paul Nankivell, Hisham Mehanna, Syed Ali Khurram, Nasir M Rajpoot
Oral epithelial dysplasia (OED) is a premalignant histopathological diagnosis given to lesions of the oral cavity. OED grading is subject to large inter/intra-rater variability, resulting in the under/over-treatment of patients. We developed a new Transformer-based pipeline to improve detection and segmentation of OED in haematoxylin and eosin (H&E) stained whole slide images (WSIs). Our model was trained on OED cases (n = 260) and controls (n = 105) collected using three different scanners, and validated on test data from three external centres in the United Kingdom and Brazil (n = 78). Our internal experiments yield a mean F1-score of 0.81 for OED segmentation, which reduced slightly to 0.71 on external testing, showing good generalisability, and gaining state-of-the-art results. This is the first externally validated study to use Transformers for segmentation in precancerous histology images. Our publicly available model shows great promise to be the first step of a fully-integrated pipeline, allowing earlier and more efficient OED diagnosis, ultimately benefiting patient outcomes.
Authors: Yang Chen, Yingwei Pan, Yehao Li, Ting Yao, Tao Mei
Recent remarkable advances in large-scale text-to-image diffusion models have inspired a significant breakthrough in text-to-3D generation, pursuing 3D content creation solely from a given text prompt. However, existing text-to-3D techniques lack a crucial ability in the creative process: interactively control and shape the synthetic 3D contents according to users' desired specifications (e.g., sketch). To alleviate this issue, we present the first attempt for text-to-3D generation conditioning on the additional hand-drawn sketch, namely Control3D, which enhances controllability for users. In particular, a 2D conditioned diffusion model (ControlNet) is remoulded to guide the learning of 3D scene parameterized as NeRF, encouraging each view of 3D scene aligned with the given text prompt and hand-drawn sketch. Moreover, we exploit a pre-trained differentiable photo-to-sketch model to directly estimate the sketch of the rendered image over synthetic 3D scene. Such estimated sketch along with each sampled view is further enforced to be geometrically consistent with the given sketch, pursuing better controllable text-to-3D generation. Through extensive experiments, we demonstrate that our proposal can generate accurate and faithful 3D scenes that align closely with the input text prompts and sketches.
Authors: Jingwen Chen, Yingwei Pan, Ting Yao, Tao Mei
Recently, the multimedia community has witnessed the rise of diffusion models trained on large-scale multi-modal data for visual content creation, particularly in the field of text-to-image generation. In this paper, we propose a new task for ``stylizing'' text-to-image models, namely text-driven stylized image generation, that further enhances editability in content creation. Given input text prompt and style image, this task aims to produce stylized images which are both semantically relevant to input text prompt and meanwhile aligned with the style image in style. To achieve this, we present a new diffusion model (ControlStyle) via upgrading a pre-trained text-to-image model with a trainable modulation network enabling more conditions of text prompts and style images. Moreover, diffusion style and content regularizations are simultaneously introduced to facilitate the learning of this modulation network with these diffusion priors, pursuing high-quality stylized text-to-image generation. Extensive experiments demonstrate the effectiveness of our ControlStyle in producing more visually pleasing and artistic results, surpassing a simple combination of text-to-image model and conventional style transfer techniques.
Authors: Haibo Yang, Yang Chen, Yingwei Pan, Ting Yao, Zhineng Chen, Tao Mei
3D content creation via text-driven stylization has played a fundamental challenge to multimedia and graphics community. Recent advances of cross-modal foundation models (e.g., CLIP) have made this problem feasible. Those approaches commonly leverage CLIP to align the holistic semantics of stylized mesh with the given text prompt. Nevertheless, it is not trivial to enable more controllable stylization of fine-grained details in 3D meshes solely based on such semantic-level cross-modal supervision. In this work, we propose a new 3DStyle-Diffusion model that triggers fine-grained stylization of 3D meshes with additional controllable appearance and geometric guidance from 2D Diffusion models. Technically, 3DStyle-Diffusion first parameterizes the texture of 3D mesh into reflectance properties and scene lighting using implicit MLP networks. Meanwhile, an accurate depth map of each sampled view is achieved conditioned on 3D mesh. Then, 3DStyle-Diffusion leverages a pre-trained controllable 2D Diffusion model to guide the learning of rendered images, encouraging the synthesized image of each view semantically aligned with text prompt and geometrically consistent with depth map. This way elegantly integrates both image rendering via implicit MLP networks and diffusion process of image synthesis in an end-to-end fashion, enabling a high-quality fine-grained stylization of 3D meshes. We also build a new dataset derived from Objaverse and the evaluation protocol for this task. Through both qualitative and quantitative experiments, we validate the capability of our 3DStyle-Diffusion. Source code and data are available at \url{https://github.com/yanghb22-fdu/3DStyle-Diffusion-Official}.
Authors: Wei-Chun Kevin Tsai, Yi-Chien Liu, Ming-Chun Yu, Chia-Ju Chou, Sui-Hing Yan, Yang-Teng Fan, Yan-Hsiang Huang, Yen-Ling Chiu, Yi-Fang Chuang, Ran-Zan Wang, Yao-Chia Shih
The Cholinergic Pathways Hyperintensities Scale (CHIPS) is a visual rating scale used to assess the extent of cholinergic white matter hyperintensities in T2-FLAIR images, serving as an indicator of dementia severity. However, the manual selection of four specific slices for rating throughout the entire brain is a time-consuming process. Our goal was to develop a deep learning-based model capable of automatically identifying the four slices relevant to CHIPS. To achieve this, we trained a 4-class slice classification model (BSCA) using the ADNI T2-FLAIR dataset (N=150) with the assistance of ResNet. Subsequently, we tested the model's performance on a local dataset (N=30). The results demonstrated the efficacy of our model, with an accuracy of 99.82% and an F1-score of 99.83%. This achievement highlights the potential impact of BSCA as an automatic screening tool, streamlining the selection of four specific T2-FLAIR slices that encompass white matter landmarks along the cholinergic pathways. Clinicians can leverage this tool to assess the risk of clinical dementia development efficiently.
Authors: Jianwei Fei, Zhihua Xia, Benedetta Tondi, Mauro Barni
In recent years, there has been significant growth in the commercial applications of generative models, licensed and distributed by model developers to users, who in turn use them to offer services. In this scenario, there is a need to track and identify the responsible user in the presence of a violation of the license agreement or any kind of malicious usage. Although there are methods enabling Generative Adversarial Networks (GANs) to include invisible watermarks in the images they produce, generating a model with a different watermark, referred to as a fingerprint, for each user is time- and resource-consuming due to the need to retrain the model to include the desired fingerprint. In this paper, we propose a retraining-free GAN fingerprinting method that allows model developers to easily generate model copies with the same functionality but different fingerprints. The generator is modified by inserting additional Personalized Normalization (PN) layers whose parameters (scaling and bias) are generated by two dedicated shallow networks (ParamGen Nets) taking the fingerprint as input. A watermark decoder is trained simultaneously to extract the fingerprint from the generated images. The proposed method can embed different fingerprints inside the GAN by just changing the input of the ParamGen Nets and performing a feedforward pass, without finetuning or retraining. The performance of the proposed method in terms of robustness against both model-level and image-level attacks is also superior to the state-of-the-art.
Authors: Yuli Wu, Weidong He, Dennis Eschweiler, Ningxin Dou, Zixin Fan, Shengli Mi, Peter Walter, Johannes Stegmaier
Modern biomedical image analysis using deep learning often encounters the challenge of limited annotated data. To overcome this issue, deep generative models can be employed to synthesize realistic biomedical images. In this regard, we propose an image synthesis method that utilizes denoising diffusion probabilistic models (DDPMs) to automatically generate retinal optical coherence tomography (OCT) images. By providing rough layer sketches, the trained DDPMs can generate realistic circumpapillary OCT images. We further find that more accurate pseudo labels can be obtained through knowledge adaptation, which greatly benefits the segmentation task. Through this, we observe a consistent improvement in layer segmentation accuracy, which is validated using various neural networks. Furthermore, we have discovered that a layer segmentation model trained solely with synthesized images can achieve comparable results to a model trained exclusively with real images. These findings demonstrate the promising potential of DDPMs in reducing the need for manual annotations of retinal OCT images.
Authors: Lei Li, Alexander Liniger, Mario Millhaeusler, Vagia Tsiminaki, Yuanyou Li, Dengxin Dai
Event cameras are gaining popularity due to their unique properties, such as their low latency and high dynamic range. One task where these benefits can be crucial is real-time object detection. However, RGB detectors still outperform event-based detectors due to the sparsity of the event data and missing visual details. In this paper, we develop a novel knowledge distillation approach to shrink the performance gap between these two modalities. To this end, we propose a cross-modality object detection distillation method that by design can focus on regions where the knowledge distillation works best. We achieve this by using an object-centric slot attention mechanism that can iteratively decouple features maps into object-centric features and corresponding pixel-features used for distillation. We evaluate our novel distillation approach on a synthetic and a real event dataset with aligned grayscale images as a teacher modality. We show that object-centric distillation allows to significantly improve the performance of the event-based student object detector, nearly halving the performance gap with respect to the teacher.
Authors: Hao-Bin Duan, Miao Wang, Jin-Chuan Shi, Xu-Chuan Chen, Yan-Pei Cao
Synthesizing photorealistic 4D human head avatars from videos is essential for VR/AR, telepresence, and video game applications. Although existing Neural Radiance Fields (NeRF)-based methods achieve high-fidelity results, the computational expense limits their use in real-time applications. To overcome this limitation, we introduce BakedAvatar, a novel representation for real-time neural head avatar synthesis, deployable in a standard polygon rasterization pipeline. Our approach extracts deformable multi-layer meshes from learned isosurfaces of the head and computes expression-, pose-, and view-dependent appearances that can be baked into static textures for efficient rasterization. We thus propose a three-stage pipeline for neural head avatar synthesis, which includes learning continuous deformation, manifold, and radiance fields, extracting layered meshes and textures, and fine-tuning texture details with differential rasterization. Experimental results demonstrate that our representation generates synthesis results of comparable quality to other state-of-the-art methods while significantly reducing the inference time required. We further showcase various head avatar synthesis results from monocular videos, including view synthesis, face reenactment, expression editing, and pose editing, all at interactive frame rates.
Authors: Lukáš Adam, Vojtěch Čermák, Kostas Papafitsoros, Lukáš Picek
This paper introduces the first public large-scale, long-span dataset with sea turtle photographs captured in the wild -- SeaTurtleID2022 (https://www.kaggle.com/datasets/wildlifedatasets/seaturtleid2022). The dataset contains 8729 photographs of 438 unique individuals collected within 13 years, making it the longest-spanned dataset for animal re-identification. All photographs include various annotations, e.g., identity, encounter timestamp, and body parts segmentation masks. Instead of standard "random" splits, the dataset allows for two realistic and ecologically motivated splits: (i) a time-aware closed-set with training, validation, and test data from different days/years, and (ii) a time-aware open-set with new unknown individuals in test and validation sets. We show that time-aware splits are essential for benchmarking re-identification methods, as random splits lead to performance overestimation. Furthermore, a baseline instance segmentation and re-identification performance over various body parts is provided. Finally, an end-to-end system for sea turtle re-identification is proposed and evaluated. The proposed system based on Hybrid Task Cascade for head instance segmentation and ArcFace-trained feature-extractor achieved an accuracy of 86.8%.
Authors: Shashanka Venkataramanan, Ewa Kijak, Laurent Amsaleg, Yannis Avrithis
Mixup refers to interpolation-based data augmentation, originally motivated as a way to go beyond empirical risk minimization (ERM). Its extensions mostly focus on the definition of interpolation and the space (input or feature) where it takes place, while the augmentation process itself is less studied. In most methods, the number of generated examples is limited to the mini-batch size and the number of examples being interpolated is limited to two (pairs), in the input space.
We make progress in this direction by introducing MultiMix, which generates an arbitrarily large number of interpolated examples beyond the mini-batch size and interpolates the entire mini-batch in the embedding space. Effectively, we sample on the entire convex hull of the mini-batch rather than along linear segments between pairs of examples.
On sequence data, we further extend to Dense MultiMix. We densely interpolate features and target labels at each spatial location and also apply the loss densely. To mitigate the lack of dense labels, we inherit labels from examples and weight interpolation factors by attention as a measure of confidence.
Overall, we increase the number of loss terms per mini-batch by orders of magnitude at little additional cost. This is only possible because of interpolating in the embedding space. We empirically show that our solutions yield significant improvement over state-of-the-art mixup methods on four different benchmarks, despite interpolation being only linear. By analyzing the embedding space, we show that the classes are more tightly clustered and uniformly spread over the embedding space, thereby explaining the improved behavior.
Authors: Simon Wiedemann, Reinhard Heckel
Cryogenic electron tomography (cryo-ET) is a technique for imaging biological samples such as viruses, cells, and proteins in 3D. A microscope collects a series of 2D projections of the sample, and the goal is to reconstruct the 3D density of the sample called the tomogram. This is difficult as the 2D projections have a missing wedge of information and are noisy. Tomograms reconstructed with conventional methods, such as filtered back-projection, suffer from the noise, and from artifacts and anisotropic resolution due to the missing wedge of information. To improve the visual quality and resolution of such tomograms, we propose a deep-learning approach for simultaneous denoising and missing wedge reconstruction called DeepDeWedge. DeepDeWedge is based on fitting a neural network to the 2D projections with a self-supervised loss inspired by noise2noise-like methods. The algorithm requires no training or ground truth data. Experiments on synthetic and real cryo-ET data show that DeepDeWedge achieves competitive performance for deep learning-based denoising and missing wedge reconstruction of cryo-ET tomograms.
Authors: Mirat Shah, Vansh Jain, Anmol Chokshi, Guruprasad Parasnis, Pramod Bide
Generative Adversarial Networks (GANs) have risen to prominence in the field of deep learning, facilitating the generation of realistic data from random noise. The effectiveness of GANs often depends on the quality of feature extraction, a critical aspect of their architecture. This paper introduces L-WaveBlock, a novel and robust feature extractor that leverages the capabilities of the Discrete Wavelet Transform (DWT) with deep learning methodologies. L-WaveBlock is catered to quicken the convergence of GAN generators while simultaneously enhancing their performance. The paper demonstrates the remarkable utility of L-WaveBlock across three datasets, a road satellite imagery dataset, the CelebA dataset and the GoPro dataset, showcasing its ability to ease feature extraction and make it more efficient. By utilizing DWT, L-WaveBlock efficiently captures the intricate details of both structural and textural details, and further partitions feature maps into orthogonal subbands across multiple scales while preserving essential information at the same time. Not only does it lead to faster convergence, but also gives competent results on every dataset by employing the L-WaveBlock. The proposed method achieves an Inception Score of 3.6959 and a Structural Similarity Index of 0.4261 on the maps dataset, a Peak Signal-to-Noise Ratio of 29.05 and a Structural Similarity Index of 0.874 on the CelebA dataset. The proposed method performs competently to the state-of-the-art for the image denoising dataset, albeit not better, but still leads to faster convergence than conventional methods. With this, L-WaveBlock emerges as a robust and efficient tool for enhancing GAN-based image generation, demonstrating superior convergence speed and competitive performance across multiple datasets for image resolution, image generation and image denoising.
Authors: Simian Luo, Yiqin Tan, Suraj Patil, Daniel Gu, Patrick von Platen, Apolinário Passos, Longbo Huang, Jian Li, Hang Zhao
Latent Consistency Models (LCMs) have achieved impressive performance in accelerating text-to-image generative tasks, producing high-quality images with minimal inference steps. LCMs are distilled from pre-trained latent diffusion models (LDMs), requiring only ~32 A100 GPU training hours. This report further extends LCMs' potential in two aspects: First, by applying LoRA distillation to Stable-Diffusion models including SD-V1.5, SSD-1B, and SDXL, we have expanded LCM's scope to larger models with significantly less memory consumption, achieving superior image generation quality. Second, we identify the LoRA parameters obtained through LCM distillation as a universal Stable-Diffusion acceleration module, named LCM-LoRA. LCM-LoRA can be directly plugged into various Stable-Diffusion fine-tuned models or LoRAs without training, thus representing a universally applicable accelerator for diverse image generation tasks. Compared with previous numerical PF-ODE solvers such as DDIM, DPM-Solver, LCM-LoRA can be viewed as a plug-in neural PF-ODE solver that possesses strong generalization abilities. Project page: https://github.com/luosiallen/latent-consistency-model.
Authors: Michael Kölle, Jonas Maurer, Philipp Altmann, Leo Sünkel, Jonas Stein, Claudia Linnhoff-Popien
Quantum computing offers the potential for superior computational capabilities, particularly for data-intensive tasks. However, the current state of quantum hardware puts heavy restrictions on input size. To address this, hybrid transfer learning solutions have been developed, merging pre-trained classical models, capable of handling extensive inputs, with variational quantum circuits. Yet, it remains unclear how much each component - classical and quantum - contributes to the model's results. We propose a novel hybrid architecture: instead of utilizing a pre-trained network for compression, we employ an autoencoder to derive a compressed version of the input data. This compressed data is then channeled through the encoder part of the autoencoder to the quantum component. We assess our model's classification capabilities against two state-of-the-art hybrid transfer learning architectures, two purely classical architectures and one quantum architecture. Their accuracy is compared across four datasets: Banknote Authentication, Breast Cancer Wisconsin, MNIST digits, and AudioMNIST. Our research suggests that classical components significantly influence classification in hybrid transfer learning, a contribution often mistakenly ascribed to the quantum element. The performance of our model aligns with that of a variational quantum circuit using amplitude embedding, positioning it as a feasible alternative.
Authors: ShengYun Peng, Seongmin Lee, Xiaojing Wang, Rajarajeswari Balasubramaniyan, Duen Horng Chau
Table structure recognition (TSR) aims to convert tabular images into a machine-readable format, where a visual encoder extracts image features and a textual decoder generates table-representing tokens. Existing approaches use classic convolutional neural network (CNN) backbones for the visual encoder and transformers for the textual decoder. However, this hybrid CNN-Transformer architecture introduces a complex visual encoder that accounts for nearly half of the total model parameters, markedly reduces both training and inference speed, and hinders the potential for self-supervised learning in TSR. In this work, we design a lightweight visual encoder for TSR without sacrificing expressive power. We discover that a convolutional stem can match classic CNN backbone performance, with a much simpler model. The convolutional stem strikes an optimal balance between two crucial factors for high-performance TSR: a higher receptive field (RF) ratio and a longer sequence length. This allows it to "see" an appropriate portion of the table and "store" the complex table structure within sufficient context length for the subsequent transformer. We conducted reproducible ablation studies and open-sourced our code at https://github.com/poloclub/tsr-convstem to enhance transparency, inspire innovations, and facilitate fair comparisons in our domain as tables are a promising modality for representation learning.
Authors: Cristina Palmero, Mikel deVelasco, Mohamed Amine Hmani, Aymen Mtibaa, Leila Ben Letaifa, Pau Buch-Cardona, Raquel Justo, Terry Amorese, Eduardo González-Fraile, Begoña Fernández-Ruanova, Jofre Tenorio-Laranga, Anna Torp Johansen, Micaela Rodrigues da Silva, Liva Jenny Martinussen, Maria Stylianou Korsnes, Gennaro Cordasco, Anna Esposito, Mounim A. El-Yacoubi, Dijana Petrovska-Delacrétaz, M. Inés Torres, Sergio Escalera
The EMPATHIC project aimed to design an emotionally expressive virtual coach capable of engaging healthy seniors to improve well-being and promote independent aging. One of the core aspects of the system is its human sensing capabilities, allowing for the perception of emotional states to provide a personalized experience. This paper outlines the development of the emotion expression recognition module of the virtual coach, encompassing data collection, annotation design, and a first methodological approach, all tailored to the project requirements. With the latter, we investigate the role of various modalities, individually and combined, for discrete emotion expression recognition in this context: speech from audio, and facial expressions, gaze, and head dynamics from video. The collected corpus includes users from Spain, France, and Norway, and was annotated separately for the audio and video channels with distinct emotional labels, allowing for a performance comparison across cultures and label types. Results confirm the informative power of the modalities studied for the emotional categories considered, with multimodal methods generally outperforming others (around 68% accuracy with audio labels and 72-74% with video labels). The findings are expected to contribute to the limited literature on emotion recognition applied to older adults in conversational human-machine interaction.
Authors: Anmol Chokshi, Vansh Jain, Rajas Bhope, Sudhir Dhage
The surge in counterfeit signatures has inflicted widespread inconveniences and formidable challenges for both individuals and organizations. This groundbreaking research paper introduces SigScatNet, an innovative solution to combat this issue by harnessing the potential of a Siamese deep learning network, bolstered by Scattering wavelets, to detect signature forgery and assess signature similarity. The Siamese Network empowers us to ascertain the authenticity of signatures through a comprehensive similarity index, enabling precise validation and comparison. Remarkably, the integration of Scattering wavelets endows our model with exceptional efficiency, rendering it light enough to operate seamlessly on cost-effective hardware systems. To validate the efficacy of our approach, extensive experimentation was conducted on two open-sourced datasets: the ICDAR SigComp Dutch dataset and the CEDAR dataset. The experimental results demonstrate the practicality and resounding success of our proposed SigScatNet, yielding an unparalleled Equal Error Rate of 3.689% with the ICDAR SigComp Dutch dataset and an astonishing 0.0578% with the CEDAR dataset. Through the implementation of SigScatNet, our research spearheads a new state-of-the-art in signature analysis in terms of EER scores and computational efficiency, offering an advanced and accessible solution for detecting forgery and quantifying signature similarities. By employing cutting-edge Siamese deep learning and Scattering wavelets, we provide a robust framework that paves the way for secure and efficient signature verification systems.
Authors: Thomas Buckley, James A. Diao, Adam Rodman, Arjun K. Manrai
Background: General-purpose large language models that utilize both text and images have not been evaluated on a diverse array of challenging medical cases.
Methods: Using 934 cases from the NEJM Image Challenge published between 2005 and 2023, we evaluated the accuracy of the recently released Generative Pre-trained Transformer 4 with Vision model (GPT-4V) compared to human respondents overall and stratified by question difficulty, image type, and skin tone. We further conducted a physician evaluation of GPT-4V on 69 NEJM clinicopathological conferences (CPCs). Analyses were conducted for models utilizing text alone, images alone, and both text and images.
Results: GPT-4V achieved an overall accuracy of 61% (95% CI, 58 to 64%) compared to 49% (95% CI, 49 to 50%) for humans. GPT-4V outperformed humans at all levels of difficulty and disagreement, skin tones, and image types; the exception was radiographic images, where performance was equivalent between GPT-4V and human respondents. Longer, more informative captions were associated with improved performance for GPT-4V but similar performance for human respondents. GPT-4V included the correct diagnosis in its differential for 80% (95% CI, 68 to 88%) of CPCs when using text alone, compared to 58% (95% CI, 45 to 70%) of CPCs when using both images and text.
Conclusions: GPT-4V outperformed human respondents on challenging medical cases and was able to synthesize information from both images and text, but performance deteriorated when images were added to highly informative text. Overall, our results suggest that multimodal AI models may be useful in medical diagnostic reasoning but that their accuracy may depend heavily on context.
Authors: Ze Yang, Sivabalan Manivasagam, Yun Chen, Jingkang Wang, Rui Hu, Raquel Urtasun
Reconstructing objects from real world data and rendering them at novel views is critical to bringing realism, diversity and scale to simulation for robotics training and testing. In this work, we present NeuSim, a novel approach that estimates accurate geometry and realistic appearance from sparse in-the-wild data captured at distance and at limited viewpoints. Towards this goal, we represent the object surface as a neural signed distance function and leverage both LiDAR and camera sensor data to reconstruct smooth and accurate geometry and normals. We model the object appearance with a robust physics-inspired reflectance representation effective for in-the-wild data. Our experiments show that NeuSim has strong view synthesis performance on challenging scenarios with sparse training views. Furthermore, we showcase composing NeuSim assets into a virtual world and generating realistic multi-sensor data for evaluating self-driving perception models.
Authors: Lakshika Rathi, Edith Tretschk, Christian Theobalt, Rishabh Dabral, Vladislav Golyanik
Existing methods for learning 3D representations are deep neural networks trained and tested on classical hardware. Quantum machine learning architectures, despite their theoretically predicted advantages in terms of speed and the representational capacity, have so far not been considered for this problem nor for tasks involving 3D data in general. This paper thus introduces the first quantum auto-encoder for 3D point clouds. Our 3D-QAE approach is fully quantum, i.e. all its data processing components are designed for quantum hardware. It is trained on collections of 3D point clouds to produce their compressed representations. Along with finding a suitable architecture, the core challenges in designing such a fully quantum model include 3D data normalisation and parameter optimisation, and we propose solutions for both these tasks. Experiments on simulated gate-based quantum hardware demonstrate that our method outperforms simple classical baselines, paving the way for a new research direction in 3D computer vision. The source code is available at https://4dqv.mpi-inf.mpg.de/QAE3D/.
Authors: Jeffrey Yunfan Liu, Yun Chen, Ze Yang, Jingkang Wang, Sivabalan Manivasagam, Raquel Urtasun
We propose a new method for realistic real-time novel-view synthesis (NVS) of large scenes. Existing neural rendering methods generate realistic results, but primarily work for small scale scenes (<50 square meters) and have difficulty at large scale (>10000 square meters). Traditional graphics-based rasterization rendering is fast for large scenes but lacks realism and requires expensive manually created assets. Our approach combines the best of both worlds by taking a moderate-quality scaffold mesh as input and learning a neural texture field and shader to model view-dependant effects to enhance realism, while still using the standard graphics pipeline for real-time rendering. Our method outperforms existing neural rendering methods, providing at least 30x faster rendering with comparable or better realism for large self-driving and drone scenes. Our work is the first to enable real-time rendering of large real-world scenes.
Authors: David Chuan-En Lin, Nikolas Martelaro
This short paper introduces a workflow for generating realistic soundscapes for visual media. In contrast to prior work, which primarily focus on matching sounds for on-screen visuals, our approach extends to suggesting sounds that may not be immediately visible but are essential to crafting a convincing and immersive auditory environment. Our key insight is leveraging the reasoning capabilities of language models, such as ChatGPT. In this paper, we describe our workflow, which includes creating a scene context, brainstorming sounds, and generating the sounds.
Authors: Daniel Bolya, Chaitanya Ryali, Judy Hoffman, Christoph Feichtenhofer
Window attention, position embeddings, and high resolution finetuning are core concepts in the modern transformer era of computer vision. However, we find that naively combining these near ubiquitous components can have a detrimental effect on performance. The issue is simple: interpolating position embeddings while using window attention is wrong. We study two state-of-the-art methods that have these three components, namely Hiera and ViTDet, and find that both do indeed suffer from this bug. To fix it, we introduce a simple absolute window position embedding strategy, which solves the bug outright in Hiera and allows us to increase both speed and performance of the model in ViTDet. We finally combine the two to obtain HieraDet, which achieves 61.7 box mAP on COCO, making it state-of-the-art for models that only use ImageNet-1k pretraining. This all stems from what is essentially a 3 line bug fix, which we name "absolute win".
Authors: Neehar Kondapaneni, Pietro Perona
Artificial intelligence (AI) systems struggle to generalize beyond their training data and abstract general properties from the specifics of the training examples. We propose a model that reproduces the apparent human ability to come up with a number sense through unsupervised everyday experience. The ability to understand and manipulate numbers and quantities emerges during childhood, but the mechanism through which humans acquire and develop this ability is still poorly understood. In particular, it is not known whether acquiring such a number sense is possible without supervision from a teacher. We explore this question through a model, assuming that the learner is able to pick and place small objects and will spontaneously engage in undirected manipulation. We assume that the learner's visual system will monitor the changing arrangements of objects in the scene and will learn to predict the effects of each action by comparing perception with the efferent signal of the motor system. We model perception using standard deep networks for feature extraction and classification. We find that, from learning the unrelated task of action prediction, an unexpected image representation emerges exhibiting regularities that foreshadow the perception and representation of numbers. These include distinct categories for the first few natural numbers, a strict ordering of the numbers, and a one-dimensional signal that correlates with numerical quantity. As a result, our model acquires the ability to estimate numerosity and subitize. Remarkably, subitization and numerosity estimation extrapolate to scenes containing many objects, far beyond the three objects used during training. We conclude that important aspects of a facility with numbers and quantities may be learned without teacher supervision.
Authors: Giuseppe Alessio D'Inverno, Monica Bianchini, Maria Lucia Sampoli, Franco Scarselli
Graph Neural Networks (GNNs) are a broad class of connectionist models for graph processing. Recent studies have shown that GNNs can approximate any function on graphs, modulo the equivalence relation on graphs defined by the Weisfeiler--Lehman (WL) test. However, these results suffer from some limitations, both because they were derived using the Stone--Weierstrass theorem -- which is existential in nature, -- and because they assume that the target function to be approximated must be continuous. Furthermore, all current results are dedicated to graph classification/regression tasks, where the GNN must produce a single output for the whole graph, while also node classification/regression problems, in which an output is returned for each node, are very common. In this paper, we propose an alternative way to demonstrate the approximation capability of GNNs that overcomes these limitations. Indeed, we show that GNNs are universal approximators in probability for node classification/regression tasks, as they can approximate any measurable function that satisfies the 1--WL equivalence on nodes. The proposed theoretical framework allows the approximation of generic discontinuous target functions and also suggests the GNN architecture that can reach a desired approximation. In addition, we provide a bound on the number of the GNN layers required to achieve the desired degree of approximation, namely $2r-1$, where $r$ is the maximum number of nodes for the graphs in the domain.
Authors: Qing Wan, Siu Wun Cheung, Yoonsuck Choe
Adjoint operators have been found to be effective in the exploration of CNN's inner workings [1]. However, the previous no-bias assumption restricted its generalization. We overcome the restriction via embedding input images into an extended normed space that includes bias in all CNN layers as part of the extended space and propose an adjoint-operator-based algorithm that maps high-level weights back to the extended input space for reconstructing an effective hypersurface. Such hypersurface can be computed for an arbitrary unit in the CNN, and we prove that this reconstructed hypersurface, when multiplied by the original input (through an inner product), will precisely replicate the output value of each unit. We show experimental results based on the CIFAR-10 and CIFAR-100 data sets where the proposed approach achieves near 0 activation value reconstruction error.
Authors: Xin Wang, Hui Guo, Shu Hu, Ming-Ching Chang, Siwei Lyu
Generative Adversarial Networks (GAN) have led to the generation of very realistic face images, which have been used in fake social media accounts and other disinformation matters that can generate profound impacts. Therefore, the corresponding GAN-face detection techniques are under active development that can examine and expose such fake faces. In this work, we aim to provide a comprehensive review of recent progress in GAN-face detection. We focus on methods that can detect face images that are generated or synthesized from GAN models. We classify the existing detection works into four categories: (1) deep learning-based, (2) physical-based, (3) physiological-based methods, and (4) evaluation and comparison against human visual performance. For each category, we summarize the key ideas and connect them with method implementations. We also discuss open problems and suggest future research directions.
Authors: Hao Quan, Xingyu Li, Weixing Chen, Qun Bai, Mingchen Zou, Ruijie Yang, Tingting Zheng, Ruiqun Qi, Xinghua Gao, Xiaoyu Cui
Based on digital pathology slice scanning technology, artificial intelligence algorithms represented by deep learning have achieved remarkable results in the field of computational pathology. Compared to other medical images, pathology images are more difficult to annotate, and thus, there is an extreme lack of available datasets for conducting supervised learning to train robust deep learning models. In this paper, we propose a self-supervised learning (SSL) model, the global contrast-masked autoencoder (GCMAE), which can train the encoder to have the ability to represent local-global features of pathological images, also significantly improve the performance of transfer learning across data sets. In this study, the ability of the GCMAE to learn migratable representations was demonstrated through extensive experiments using a total of three different disease-specific hematoxylin and eosin (HE)-stained pathology datasets: Camelyon16, NCTCRC and BreakHis. In addition, this study designed an effective automated pathology diagnosis process based on the GCMAE for clinical applications. The source code of this paper is publicly available at https://github.com/StarUniversus/gcmae.
Authors: Iris Yan
Cancer is the second leading cause of death in the world. Diagnosing cancer early on can save many lives. Pathologists have to look at tissue microarray (TMA) images manually to identify tumors, which can be time-consuming, inconsistent and subjective. Existing automatic algorithms either have not achieved the accuracy level of a pathologist or require substantial human involvements. A major challenge is that TMA images with different shapes, sizes, and locations can have the same score. Learning staining patterns in TMA images requires a huge number of images, which are severely limited due to privacy and regulation concerns in medical organizations. TMA images from different cancer types may share certain common characteristics, but combining them directly harms the accuracy due to heterogeneity in their staining patterns. Transfer learning is an emerging learning paradigm that allows borrowing strength from similar problems. However, existing approaches typically require a large sample from similar learning problems, while TMA images of different cancer types are often available in small sample size and further existing algorithms are limited to transfer learning from one similar problem. We propose a new transfer learning algorithm that could learn from multiple related problems, where each problem has a small sample and can have a substantially different distribution from the original one. The proposed algorithm has made it possible to break the critical accuracy barrier (the 75% accuracy level of pathologists), with a reported accuracy of 75.9% on breast cancer TMA images from the Stanford Tissue Microarray Database. It is supported by recent developments in transfer learning theory and empirical evidence in clustering technology. This will allow pathologists to confidently adopt automatic algorithms in recognizing tumors consistently with a higher accuracy in real time.
Authors: Xin Wang, Jing-Ke Yan, Jing-Ye Cai, Jian-Hua Deng, Qin Qin, Yao Cheng
Single-image super-resolution (SISR) typically focuses on restoring various degraded low-resolution (LR) images to a single high-resolution (HR) image. However, during SISR tasks, it is often challenging for models to simultaneously maintain high quality and rapid sampling while preserving diversity in details and texture features. This challenge can lead to issues such as model collapse, lack of rich details and texture features in the reconstructed HR images, and excessive time consumption for model sampling. To address these problems, this paper proposes a Latent Feature-oriented Diffusion Probability Model (LDDPM). First, we designed a conditional encoder capable of effectively encoding LR images, reducing the solution space for model image reconstruction and thereby improving the quality of the reconstructed images. We then employed a normalized flow and multimodal adversarial training, learning from complex multimodal distributions, to model the denoising distribution. Doing so boosts the generative modeling capabilities within a minimal number of sampling steps. Experimental comparisons of our proposed model with existing SISR methods on mainstream datasets demonstrate that our model reconstructs more realistic HR images and achieves better performance on multiple evaluation metrics, providing a fresh perspective for tackling SISR tasks.
Authors: Jiawen Li, Yunqian Huang, Sheng Song, Hongbo Chen, Junni Shi, Duo Xu, Haibin Zhang, Man Chen, Rui Zheng
The objective of this study is to develop a deep-learning based detection and diagnosis technique for carotid atherosclerosis using a portable freehand 3D ultrasound (US) imaging system. A total of 127 3D carotid artery scans were acquired using a portable 3D US system which consisted of a handheld US scanner and an electromagnetic tracking system. A U-Net segmentation network was firstly applied to extract the carotid artery on 2D transverse frame, then a novel 3D reconstruction algorithm using fast dot projection (FDP) method with position regularization was proposed to reconstruct the carotid artery volume. Furthermore, a convolutional neural network was used to classify healthy and diseased cases qualitatively. 3D volume analysis methods including longitudinal image acquisition and stenosis grade measurement were developed to obtain the clinical metrics quantitatively. The proposed system achieved sensitivity of 0.714, specificity of 0.851 and accuracy of 0.803 respectively for diagnosis of carotid atherosclerosis. The automatically measured stenosis grade illustrated good correlation (r=0.762) with the experienced expert measurement. The developed technique based on 3D US imaging can be applied to the automatic diagnosis of carotid atherosclerosis. The proposed deep-learning based technique was specially designed for a portable 3D freehand US system, which can provide more convenient carotid atherosclerosis examination and decrease the dependence on clinician's experience.
Authors: R T Akash Guna, Raul Benitez, O K Sikha
We propose a novel method that trains a conditional Generative Adversarial Network (GAN) to generate visual interpretations of a Convolutional Neural Network (CNN). To comprehend a CNN, the GAN is trained with information on how the CNN processes an image when making predictions. Supplying that information has two main challenges: how to represent this information in a form that is feedable to the GANs and how to effectively feed the representation to the GAN. To address these issues, we developed a suitable representation of CNN architectures by cumulatively averaging intermediate interpretation maps. We also propose two alternative approaches to feed the representations to the GAN and to choose an effective training strategy. Our approach learned the general aspects of CNNs and was agnostic to datasets and CNN architectures. The study includes both qualitative and quantitative evaluations and compares the proposed GANs with state-of-the-art approaches. We found that the initial layers of CNNs and final layers are equally crucial for interpreting CNNs upon interpreting the proposed GAN. We believe training a GAN to interpret CNNs would open doors for improved interpretations by leveraging fast-paced deep learning advancements. The code used for experimentation is publicly available at https://github.com/Akash-guna/Explain-CNN-With-GANS
Authors: Zunzhi You, Daochang Liu, Bohyung Han, Chang Xu
Recent advancements in masked image modeling (MIM) have made it a prevailing framework for self-supervised visual representation learning. The MIM pretrained models, like most deep neural network methods, remain vulnerable to adversarial attacks, limiting their practical application, and this issue has received little research attention. In this paper, we investigate how this powerful self-supervised learning paradigm can provide adversarial robustness to downstream classifiers. During the exploration, we find that noisy image modeling (NIM), a simple variant of MIM that adopts denoising as the pre-text task, reconstructs noisy images surprisingly well despite severe corruption. Motivated by this observation, we propose an adversarial defense method, referred to as De^3, by exploiting the pretrained decoder for denoising. Through De^3, NIM is able to enhance adversarial robustness beyond providing pretrained features. Furthermore, we incorporate a simple modification, sampling the noise scale hyperparameter from random distributions, and enable the defense to achieve a better and tunable trade-off between accuracy and robustness. Experimental results demonstrate that, in terms of adversarial robustness, NIM is superior to MIM thanks to its effective denoising capability. Moreover, the defense provided by NIM achieves performance on par with adversarial training while offering the extra tunability advantage. Source code and models are available at https://github.com/youzunzhi/NIM-AdvDef.
Authors: Nikolas Adaloglou, Felix Michels, Tim Kaiser, Markus Kollmann
We present a comprehensive experimental study on pretrained feature extractors for visual out-of-distribution (OOD) detection, focusing on adapting contrastive language-image pretrained (CLIP) models. Without fine-tuning on the training data, we are able to establish a positive correlation ($R^2\geq0.92$) between in-distribution classification and unsupervised OOD detection for CLIP models in $4$ benchmarks. We further propose a new simple and scalable method called \textit{pseudo-label probing} (PLP) that adapts vision-language models for OOD detection. Given a set of label names of the training set, PLP trains a linear layer using the pseudo-labels derived from the text encoder of CLIP. To test the OOD detection robustness of pretrained models, we develop a novel feature-based adversarial OOD data manipulation approach to create adversarial samples. Intriguingly, we show that (i) PLP outperforms the previous state-of-the-art \citep{ming2022mcm} on all $5$ large-scale benchmarks based on ImageNet, specifically by an average AUROC gain of 3.4\% using the largest CLIP model (ViT-G), (ii) we show that linear probing outperforms fine-tuning by large margins for CLIP architectures (i.e. CLIP ViT-H achieves a mean gain of 7.3\% AUROC on average on all ImageNet-based benchmarks), and (iii) billion-parameter CLIP models still fail at detecting adversarially manipulated OOD images. The code and adversarially created datasets will be made publicly available.
Authors: Bjoern Michele, Alexandre Boulch, Gilles Puy, Tuan-Hung Vu, Renaud Marlet, Nicolas Courty
Learning models on one labeled dataset that generalize well on another domain is a difficult task, as several shifts might happen between the data domains. This is notably the case for lidar data, for which models can exhibit large performance discrepancies due for instance to different lidar patterns or changes in acquisition conditions. This paper addresses the corresponding Unsupervised Domain Adaptation (UDA) task for semantic segmentation. To mitigate this problem, we introduce an unsupervised auxiliary task of learning an implicit underlying surface representation simultaneously on source and target data. As both domains share the same latent representation, the model is forced to accommodate discrepancies between the two sources of data. This novel strategy differs from classical minimization of statistical divergences or lidar-specific domain adaptation techniques. Our experiments demonstrate that our method achieves a better performance than the current state of the art, both in real-to-real and synthetic-to-real scenarios.
Authors: Simiao Ren, Francesco Luzi, Saad Lahrichi, Kaleb Kassaw, Leslie M. Collins, Kyle Bradbury, Jordan M. Malof
Recently, the first foundation model developed specifically for image segmentation tasks was developed, termed the "Segment Anything Model" (SAM). SAM can segment objects in input imagery based on cheap input prompts, such as one (or more) points, a bounding box, or a mask. The authors examined the \textit{zero-shot} image segmentation accuracy of SAM on a large number of vision benchmark tasks and found that SAM usually achieved recognition accuracy similar to, or sometimes exceeding, vision models that had been trained on the target tasks. The impressive generalization of SAM for segmentation has major implications for vision researchers working on natural imagery. In this work, we examine whether SAM's performance extends to overhead imagery problems and help guide the community's response to its development. We examine SAM's performance on a set of diverse and widely studied benchmark tasks. We find that SAM does often generalize well to overhead imagery, although it fails in some cases due to the unique characteristics of overhead imagery and its common target objects. We report on these unique systematic failure cases for remote sensing imagery that may comprise useful future research for the community.
Authors: Fangwen Wu, Jingxuan He, Yufei Yin, Yanbin Hao, Gang Huang, Lechao Cheng
This study introduces an efficacious approach, Masked Collaborative Contrast (MCC), to highlight semantic regions in weakly supervised semantic segmentation. MCC adroitly draws inspiration from masked image modeling and contrastive learning to devise a novel framework that induces keys to contract toward semantic regions. Unlike prevalent techniques that directly eradicate patch regions in the input image when generating masks, we scrutinize the neighborhood relations of patch tokens by exploring masks considering keys on the affinity matrix. Moreover, we generate positive and negative samples in contrastive learning by utilizing the masked local output and contrasting it with the global output. Elaborate experiments on commonly employed datasets evidences that the proposed MCC mechanism effectively aligns global and local perspectives within the image, attaining impressive performance. The source code is available at \url{https://github.com/fwu11/MCC}.
Authors: Vaishnavi Himakunthala, Andy Ouyang, Daniel Rose, Ryan He, Alex Mei, Yujie Lu, Chinmay Sonar, Michael Saxon, William Yang Wang
Despite exciting recent results showing vision-language systems' capacity to reason about images using natural language, their capacity for video reasoning remains under-explored. We motivate framing video reasoning as the sequential understanding of a small number of keyframes, thereby leveraging the power and robustness of vision-language while alleviating the computational complexities of processing videos. To evaluate this novel application, we introduce VIP, an inference-time challenge dataset designed to explore models' reasoning capabilities through video chain-of-thought. Inspired by visually descriptive scene plays, we propose two formats for keyframe description: unstructured dense captions and structured scene descriptions that identify the focus, action, mood, objects, and setting (FAMOuS) of the keyframe. To evaluate video reasoning, we propose two tasks: Video Infilling and Video Prediction, which test abilities to generate multiple intermediate keyframes and predict future keyframes, respectively. We benchmark GPT-4, GPT-3, and VICUNA on VIP, demonstrate the performance gap in these complex video reasoning tasks, and encourage future work to prioritize language models for efficient and generalized video reasoning.
Authors: Gabriel Kasmi, Laurent Dubus, Yves-Marie Saint Drenan, Philippe Blanc
Neural networks have shown remarkable performance in computer vision, but their deployment in numerous scientific and technical fields is challenging due to their black-box nature. Scientists and practitioners need to evaluate the reliability of a decision, i.e., to know simultaneously if a model relies on the relevant features and whether these features are robust to image corruptions. Existing attribution methods aim to provide human-understandable explanations by highlighting important regions in the image domain, but fail to fully characterize a decision process's reliability. To bridge this gap, we introduce the Wavelet sCale Attribution Method (WCAM), a generalization of attribution from the pixel domain to the space-scale domain using wavelet transforms. Attribution in the wavelet domain reveals where and on what scales the model focuses, thus enabling us to assess whether a decision is reliable. Our code is accessible here: \url{https://github.com/gabrielkasmi/spectral-attribution}.
Authors: Aleksandar Stanić, Anand Gopalakrishnan, Kazuki Irie, Jürgen Schmidhuber
Current state-of-the-art object-centric models use slots and attention-based routing for binding. However, this class of models has several conceptual limitations: the number of slots is hardwired; all slots have equal capacity; training has high computational cost; there are no object-level relational factors within slots. Synchrony-based models in principle can address these limitations by using complex-valued activations which store binding information in their phase components. However, working examples of such synchrony-based models have been developed only very recently, and are still limited to toy grayscale datasets and simultaneous storage of less than three objects in practice. Here we introduce architectural modifications and a novel contrastive learning method that greatly improve the state-of-the-art synchrony-based model. For the first time, we obtain a class of synchrony-based models capable of discovering objects in an unsupervised manner in multi-object color datasets and simultaneously representing more than three objects.
Authors: Wenhui Zhu, Peijie Qiu, Xin Li, Natasha Lepore, Oana M. Dumitrascu, Yalin Wang
Over the past few decades, convolutional neural networks (CNNs) have been at the forefront of the detection and tracking of various retinal diseases (RD). Despite their success, the emergence of vision transformers (ViT) in the 2020s has shifted the trajectory of RD model development. The leading-edge performance of ViT-based models in RD can be largely credited to their scalability - their ability to improve as more parameters are added. As a result, ViT-based models tend to outshine traditional CNNs in RD applications, albeit at the cost of increased data and computational demands. ViTs also differ from CNNs in their approach to processing images, working with patches rather than local regions, which can complicate the precise identification of small, variably presented lesions in RD. In our study, we revisited and updated the architecture of a CNN model, specifically MobileNet, to enhance its utility in RD diagnostics. We found that an optimized MobileNet, through selective modifications, can surpass ViT-based models in various RD benchmarks, including diabetic retinopathy grading, detection of multiple fundus diseases, and classification of diabetic macular edema. Our software package is available at https://github.com/Retinal-Research/NN-MOBILENET
Authors: Colton R. Crum, Aidan Boyd, Kevin Bowyer, Adam Czajka
Machine learning models have shown increased accuracy in classification tasks when the training process incorporates human perceptual information. However, a challenge in training human-guided models is the cost associated with collecting image annotations for human salience. Collecting annotation data for all images in a large training set can be prohibitively expensive. In this work, we utilize "teacher" models (trained on a small amount of human-annotated data) to annotate additional data by means of teacher models' saliency maps. Then, "student" models are trained using the larger amount of annotated training data. This approach makes it possible to supplement a limited number of human-supplied annotations with an arbitrarily large number of model-generated image annotations. We compare the accuracy achieved by our teacher-student training paradigm with (1) training using all available human salience annotations, and (2) using all available training data without human salience annotations. We use synthetic face detection and fake iris detection as example challenging problems, and report results across four model architectures (DenseNet, ResNet, Xception, and Inception), and two saliency estimation methods (CAM and RISE). Results show that our teacher-student training paradigm results in models that significantly exceed the performance of both baselines, demonstrating that our approach can usefully leverage a small amount of human annotations to generate salience maps for an arbitrary amount of additional training data.
Authors: Ziyi Xi, Wenmin Huang, Kangkang Wei, Weiqi Luo, Peijia Zheng
With the rapid evolution of AI Generated Content (AIGC), forged images produced through this technology are inherently more deceptive and require less human intervention compared to traditional Computer-generated Graphics (CG). However, owing to the disparities between CG and AIGC, conventional CG detection methods tend to be inadequate in identifying AIGC-produced images. To address this issue, our research concentrates on the text-to-image generation process in AIGC. Initially, we first assemble two text-to-image databases utilizing two distinct AI systems, DALLE2 and DreamStudio. Aiming to holistically capture the inherent anomalies produced by AIGC, we develope a robust dual-stream network comprised of a residual stream and a content stream. The former employs the Spatial Rich Model (SRM) to meticulously extract various texture information from images, while the latter seeks to capture additional forged traces in low frequency, thereby extracting complementary information that the residual stream may overlook. To enhance the information exchange between these two streams, we incorporate a cross multi-head attention mechanism. Numerous comparative experiments are performed on both databases, and the results show that our detection method consistently outperforms traditional CG detection techniques across a range of image resolutions. Moreover, our method exhibits superior performance through a series of robustness tests and cross-database experiments. When applied to widely recognized traditional CG benchmarks such as SPL2018 and DsTok, our approach significantly exceeds the capabilities of other existing methods in the field of CG detection.
Authors: Michael Tschannen, Manoj Kumar, Andreas Steiner, Xiaohua Zhai, Neil Houlsby, Lucas Beyer
Contrastive pretraining on image-text pairs from the web is one of the most popular large-scale pretraining strategies for vision backbones, especially in the context of large multimodal models. At the same time, image captioning on this type of data is commonly considered an inferior pretraining strategy. In this paper, we perform a fair comparison of these two pretraining strategies, carefully matching training data, compute, and model capacity. Using a standard encoder-decoder transformer, we find that captioning alone is surprisingly effective: on classification tasks, captioning produces vision encoders competitive with contrastively pretrained encoders, while surpassing them on vision & language tasks. We further analyze the effect of the model architecture and scale, as well as the pretraining data on the representation quality, and find that captioning exhibits the same or better scaling behavior along these axes. Overall our results show that plain image captioning is a more powerful pretraining strategy than was previously believed.
Authors: Pum Jun Kim, Yoojin Jang, Jisu Kim, Jaejun Yoo
We propose a robust and reliable evaluation metric for generative models by introducing topological and statistical treatments for rigorous support estimation. Existing metrics, such as Inception Score (IS), Frechet Inception Distance (FID), and the variants of Precision and Recall (P&R), heavily rely on supports that are estimated from sample features. However, the reliability of their estimation has not been seriously discussed (and overlooked) even though the quality of the evaluation entirely depends on it. In this paper, we propose Topological Precision and Recall (TopP&R, pronounced 'topper'), which provides a systematic approach to estimating supports, retaining only topologically and statistically important features with a certain level of confidence. This not only makes TopP&R strong for noisy features, but also provides statistical consistency. Our theoretical and experimental results show that TopP&R is robust to outliers and non-independent and identically distributed (Non-IID) perturbations, while accurately capturing the true trend of change in samples. To the best of our knowledge, this is the first evaluation metric focused on the robust estimation of the support and provides its statistical consistency under noise.
Authors: Yelysei Bondarenko, Markus Nagel, Tijmen Blankevoort
Transformer models have been widely adopted in various domains over the last years, and especially large language models have advanced the field of AI significantly. Due to their size, the capability of these networks has increased tremendously, but this has come at the cost of a significant increase in necessary compute. Quantization is one of the most effective ways to reduce the computational time and memory consumption of neural networks. Many studies have shown, however, that modern transformer models tend to learn strong outliers in their activations, making them difficult to quantize. To retain acceptable performance, the existence of these outliers requires activations to be in higher bitwidth or the use of different numeric formats, extra fine-tuning, or other workarounds. We show that strong outliers are related to very specific behavior of attention heads that try to learn a "no-op" or just a partial update of the residual. To achieve the exact zeros needed in the attention matrix for a no-update, the input to the softmax is pushed to be larger and larger during training, causing outliers in other parts of the network. Based on these observations, we propose two simple (independent) modifications to the attention mechanism - clipped softmax and gated attention. We empirically show that models pre-trained using our methods learn significantly smaller outliers while maintaining and sometimes even improving the floating-point task performance. This enables us to quantize transformers to full INT8 quantization of the activations without any additional effort. We demonstrate the effectiveness of our methods on both language models (BERT, OPT) and vision transformers.
Authors: Silas Santiago Lopes Pereira, José Everardo Bessa Maia
Occlusion and clutter are two scene states that make it difficult to detect anomalies in surveillance video. Furthermore, anomaly events are rare and, as a consequence, class imbalance and lack of labeled anomaly data are also key features of this task. Therefore, weakly supervised methods are heavily researched for this application. In this paper, we tackle these typical problems of anomaly detection in surveillance video by combining Multiple Instance Learning (MIL) to deal with the lack of labels and Multiple Camera Views (MC) to reduce occlusion and clutter effects. In the resulting MC-MIL algorithm we apply a multiple camera combined loss function to train a regression network with Sultani's MIL ranking function. To evaluate the MC-MIL algorithm first proposed here, the multiple camera PETS-2009 benchmark dataset was re-labeled for the anomaly detection task from multiple camera views. The result shows a significant performance improvement in F1 score compared to the single-camera configuration.
Authors: Luke Ditria, Tom Drummond
Generative models have seen an explosion in popularity with the release of huge generative Diffusion models like Midjourney and Stable Diffusion to the public. Because of this new ease of access, questions surrounding the automated collection of data and issues regarding content ownership have started to build. In this paper we present new work which aims to provide ways of protecting content when shared to the public. We show that a generative Diffusion model trained on data that has been imperceptibly watermarked will generate new images with these watermarks present. We further show that if a given watermark is correlated with a certain feature of the training data, the generated images will also have this correlation. Using statistical tests we show that we are able to determine whether a model has been trained on marked data, and what data was marked. As a result our system offers a solution to protect intellectual property when sharing content online.
Authors: Loïc Le Bescond, Maria Vakalopoulou, Stergios Christodoulidis, Fabrice André, Hugues Talbot
The deployment of machine learning solutions in real-world scenarios often involves addressing the challenge of out-of-distribution (OOD) detection. While significant efforts have been devoted to OOD detection in classical supervised settings, the context of weakly supervised learning, particularly the Multiple Instance Learning (MIL) framework, remains under-explored. In this study, we tackle this challenge by adapting post-hoc OOD detection methods to the MIL setting while introducing a novel benchmark specifically designed to assess OOD detection performance in weakly supervised scenarios. Across extensive experiments based on diverse public datasets, KNN emerges as the best-performing method overall. However, it exhibits significant shortcomings on some datasets, emphasizing the complexity of this under-explored and challenging topic. Our findings shed light on the complex nature of OOD detection under the MIL framework, emphasizing the importance of developing novel, robust, and reliable methods that can generalize effectively in a weakly supervised context. The code for the paper is available here: https://github.com/loic-lb/OOD_MIL.
Authors: James K Ruffle, Robert J Gray, Samia Mohinta, Guilherme Pombo, Chaitanya Kaul, Harpreet Hyare, Geraint Rees, Parashkev Nachev
Our knowledge of the organisation of the human brain at the population-level is yet to translate into power to predict functional differences at the individual-level, limiting clinical applications, and casting doubt on the generalisability of inferred mechanisms. It remains unknown whether the difficulty arises from the absence of individuating biological patterns within the brain, or from limited power to access them with the models and compute at our disposal. Here we comprehensively investigate the resolvability of such patterns with data and compute at unprecedented scale. Across 23810 unique participants from UK Biobank, we systematically evaluate the predictability of 25 individual biological characteristics, from all available combinations of structural and functional neuroimaging data. Over 4526 GPU*hours of computation, we train, optimize, and evaluate out-of-sample 700 individual predictive models, including multilayer perceptrons of demographic, psychological, serological, chronic morbidity, and functional connectivity characteristics, and both uni- and multi-modal 3D convolutional neural network models of macro- and micro-structural brain imaging. We find a marked discrepancy between the high predictability of sex (balanced accuracy 99.7%), age (mean absolute error 2.048 years, R2 0.859), and weight (mean absolute error 2.609Kg, R2 0.625), for which we set new state-of-the-art performance, and the surprisingly low predictability of other characteristics. Neither structural nor functional imaging predicted individual psychology better than the coincidence of common chronic morbidity (p<0.05). Serology predicted common morbidity (p<0.05) and was best predicted by it (p<0.001), followed by structural neuroimaging (p<0.05). Our findings suggest either more informative imaging or more powerful models will be needed to decipher individual level characteristics from the brain.
Authors: Gabriel Kasmi, Laurent Dubus, Yves-Marie Saint-Drenan, Philippe Blanc
Photovoltaic (PV) energy is crucial for the decarbonization of energy systems. Due to the lack of centralized data, remote sensing of rooftop PV installations is the best option to monitor the evolution of the rooftop PV installed fleet at a regional scale. However, current techniques lack reliability and are notably sensitive to shifts in the acquisition conditions. To overcome this, we leverage the wavelet scale attribution method (WCAM), which decomposes a model's prediction in the space-scale domain. The WCAM enables us to assess on which scales the representation of a PV model rests and provides insights to derive methods that improve the robustness to acquisition conditions, thus increasing trust in deep learning systems to encourage their use for the safe integration of clean energy in electric systems.
Authors: Shiyi Tang, Shu Zhang, Yini Fang
Small object detection has been a challenging problem in the field of object detection. There has been some works that proposes improvements for this task, such as adding several attention blocks or changing the whole structure of feature fusion networks. However, the computation cost of these models is large, which makes deploying a real-time object detection system unfeasible, while leaving room for improvement. To this end, an improved YOLOv5 model: HIC-YOLOv5 is proposed to address the aforementioned problems. Firstly, an additional prediction head specific to small objects is added to provide a higher-resolution feature map for better prediction. Secondly, an involution block is adopted between the backbone and neck to increase channel information of the feature map. Moreover, an attention mechanism named CBAM is applied at the end of the backbone, thus not only decreasing the computation cost compared with previous works but also emphasizing the important information in both channel and spatial domain. Our result shows that HIC-YOLOv5 has improved mAP@[.5:.95] by 6.42% and mAP@0.5 by 9.38% on VisDrone-2019-DET dataset.
Authors: Xinyu Zhang, Li Wang, Jian Chen, Cheng Fang, Lei Yang, Ziying Song, Guangqi Yang, Yichen Wang, Xiaofei Zhang, Jun Li, Zhiwei Li, Qingshan Yang, Zhenlin Zhang, Shuzhi Sam Ge
Radar has stronger adaptability in adverse scenarios for autonomous driving environmental perception compared to widely adopted cameras and LiDARs. Compared with commonly used 3D radars, the latest 4D radars have precise vertical resolution and higher point cloud density, making it a highly promising sensor for autonomous driving in complex environmental perception. However, due to the much higher noise than LiDAR, manufacturers choose different filtering strategies, resulting in an inverse ratio between noise level and point cloud density. There is still a lack of comparative analysis on which method is beneficial for deep learning-based perception algorithms in autonomous driving. One of the main reasons is that current datasets only adopt one type of 4D radar, making it difficult to compare different 4D radars in the same scene. Therefore, in this paper, we introduce a novel large-scale multi-modal dataset featuring, for the first time, two types of 4D radars captured simultaneously. This dataset enables further research into effective 4D radar perception algorithms.Our dataset consists of 151 consecutive series, most of which last 20 seconds and contain 10,007 meticulously synchronized and annotated frames. Moreover, our dataset captures a variety of challenging driving scenarios, including many road conditions, weather conditions, nighttime and daytime with different lighting intensities and periods. Our dataset annotates consecutive frames, which can be applied to 3D object detection and tracking, and also supports the study of multi-modal tasks. We experimentally validate our dataset, providing valuable results for studying different types of 4D radars. This dataset is released on https://github.com/adept-thu/Dual-Radar.
Authors: Jiaxin Wei, Stefan Leutenegger, Laurent Kneip
Perspective-$n$-Point (P$n$P) stands as a fundamental algorithm for pose estimation in various applications. In this paper, we present a new approach to the P$n$P problem with relaxed constraints, eliminating the need for precise 3D coordinates or complete calibration data. We refer to it as AP$n$P due to its ability to handle unknown anisotropic scaling factors of 3D coordinates or alternatively two distinct focal lengths in addition to the conventional rigid transformation. Through algebraic manipulations and a novel parametrization, both cases are brought into similar forms that distinguish themselves primarily by the order of a rotation and an anisotropic scaling operation. AP$n$P then boils down to one unique polynomial problem, which is solved by the Gr\"obner basis approach. Experimental results on both simulated and real datasets demonstrate the effectiveness of AP$n$P as a more flexible and practical solution to camera pose estimation. Code: https://github.com/goldoak/APnP.
Authors: Xiaoxiao Long, Yuan-Chen Guo, Cheng Lin, Yuan Liu, Zhiyang Dou, Lingjie Liu, Yuexin Ma, Song-Hai Zhang, Marc Habermann, Christian Theobalt, Wenping Wang
In this work, we introduce Wonder3D, a novel method for efficiently generating high-fidelity textured meshes from single-view images.Recent methods based on Score Distillation Sampling (SDS) have shown the potential to recover 3D geometry from 2D diffusion priors, but they typically suffer from time-consuming per-shape optimization and inconsistent geometry. In contrast, certain works directly produce 3D information via fast network inferences, but their results are often of low quality and lack geometric details. To holistically improve the quality, consistency, and efficiency of image-to-3D tasks, we propose a cross-domain diffusion model that generates multi-view normal maps and the corresponding color images. To ensure consistency, we employ a multi-view cross-domain attention mechanism that facilitates information exchange across views and modalities. Lastly, we introduce a geometry-aware normal fusion algorithm that extracts high-quality surfaces from the multi-view 2D representations. Our extensive evaluations demonstrate that our method achieves high-quality reconstruction results, robust generalization, and reasonably good efficiency compared to prior works.
Authors: Hao Li, Quanwei Liu, Jianan Liu, Xiling Liu, Yanni Dong, Tao Huang, Zhihan Lv
High-resolution (HR) magnetic resonance imaging (MRI) is crucial for enhancing diagnostic accuracy in clinical settings. Nonetheless, the inherent limitation of MRI resolution restricts its widespread applicability. Deep learning-based image super-resolution (SR) methods exhibit promise in improving MRI resolution without additional cost. However, these methods frequently require a substantial number of HR MRI images for training, which can be challenging to acquire. In this paper, we propose an unpaired MRI SR approach that employs self-supervised contrastive learning to enhance SR performance with limited training data. Our approach leverages both authentic HR images and synthetically generated SR images to construct positive and negative sample pairs, thus facilitating the learning of discriminative features. Empirical results presented in this study underscore significant enhancements in the peak signal-to-noise ratio and structural similarity index, even when a paucity of HR images is available. These findings accentuate the potential of our approach in addressing the challenge of limited training data, thereby contributing to the advancement of high-resolution MRI in clinical applications.
Authors: Olivier Rukundo, Andrea Behanova, Riccardo De Feo, Seppo Ronkko, Joni Oja, Jussi Tohka
Regular monitoring of the primary particles and purity profiles of a drug product during development and manufacturing processes is essential for manufacturers to avoid product variability and contamination. Transmission electron microscopy (TEM) imaging helps manufacturers predict how changes affect particle characteristics and purity for virus-based gene therapy vector products and intermediates. Since intact particles can characterize efficacious products, it is beneficial to automate the detection of intact adenovirus against a non-intact-viral background mixed with debris, broken, and artefact particles. In the presence of such particles, detecting intact adenoviruses becomes more challenging. To overcome the challenge, due to such a presence, we developed a software tool for semi-automatic annotation and segmentation of adenoviruses and a software tool for automatic segmentation and detection of intact adenoviruses in TEM imaging systems. The developed semi-automatic tool exploited conventional image analysis techniques while the automatic tool was built based on convolutional neural networks and image analysis techniques. Our quantitative and qualitative evaluations showed outstanding true positive detection rates compared to false positive and negative rates where adenoviruses were nicely detected without mistaking them for real debris, broken adenoviruses, and/or staining artefacts.
Authors: Md Azim Khan
Several visual tasks, such as pedestrian detection and image-to-image translation, are challenging to accomplish in low light using RGB images. Heat variation of objects in thermal images can be used to overcome this. In this work, an end-to-end framework, which consists of a generative network and a detector network, is proposed to translate RGB image into Thermal ones and compare generated thermal images with real data. We have collected images from two different locations using the Parrot Anafi Thermal drone. After that, we created a two-stream network, preprocessed, augmented, the image data, and trained the generator and discriminator models from scratch. The findings demonstrate that it is feasible to translate RGB training data to thermal data using GAN. As a result, thermal data can now be produced more quickly and affordably, which is useful for security and surveillance applications.
Authors: Diego Hidalgo-Carvajal, Hanzhi Chen, Gemma C. Bettelani, Jaesug Jung, Melissa Zavaglia, Laura Busse, Abdeldjallil Naceri, Stefan Leutenegger, Sami Haddadin
The progressive prevalence of robots in human-suited environments has given rise to a myriad of object manipulation techniques, in which dexterity plays a paramount role. It is well-established that humans exhibit extraordinary dexterity when handling objects. Such dexterity seems to derive from a robust understanding of object properties (such as weight, size, and shape), as well as a remarkable capacity to interact with them. Hand postures commonly demonstrate the influence of specific regions on objects that need to be grasped, especially when objects are partially visible. In this work, we leverage human-like object understanding by reconstructing and completing their full geometry from partial observations, and manipulating them using a 7-DoF anthropomorphic robot hand. Our approach has significantly improved the grasping success rates of baselines with only partial reconstruction by nearly 30% and achieved over 150 successful grasps with three different object categories. This demonstrates our approach's consistent ability to predict and execute grasping postures based on the completed object shapes from various directions and positions in real-world scenarios. Our work opens up new possibilities for enhancing robotic applications that require precise grasping and manipulation skills of real-world reconstructed objects.
Authors: Zhenzhong Cao
Loop closure, as one of the crucial components in SLAM, plays an essential role in correcting the accumulated errors. Traditional appearance-based methods, such as bag-of-words models, are often limited by local 2D features and the volume of training data, making them less versatile and robust in real-world scenarios, leading to missed detections or false positives detections in loop closure. To address these issues, we first propose a object-level data association method based on multi-level verification, which can associate 2D semantic features of current frame with 3D objects landmarks of map. Next, taking advantage of these association relations, we introduce a semantic loop closure method based on quadric-level object map topology, which represents scenes through the topological graph of objects and achieves accurate loop closure at a wide field of view by comparing differences in the topological graphs. Finally, we integrate these two methods into a complete object-aware SLAM system. Qualitative experiments and ablation studies demonstrate the effectiveness and robustness of the proposed object-level data association algorithm. Quantitative experiments show that our semantic loop closure method outperforms existing state-of-the-art methods in terms of precision, recall and localization accuracy metrics.
Authors: Qitao Zhao, Ce Zheng, Mengyuan Liu, Chen Chen
The dominant paradigm in 3D human pose estimation that lifts a 2D pose sequence to 3D heavily relies on long-term temporal clues (i.e., using a daunting number of video frames) for improved accuracy, which incurs performance saturation, intractable computation and the non-causal problem. This can be attributed to their inherent inability to perceive spatial context as plain 2D joint coordinates carry no visual cues. To address this issue, we propose a straightforward yet powerful solution: leveraging the readily available intermediate visual representations produced by off-the-shelf (pre-trained) 2D pose detectors -- no finetuning on the 3D task is even needed. The key observation is that, while the pose detector learns to localize 2D joints, such representations (e.g., feature maps) implicitly encode the joint-centric spatial context thanks to the regional operations in backbone networks. We design a simple baseline named Context-Aware PoseFormer to showcase its effectiveness. Without access to any temporal information, the proposed method significantly outperforms its context-agnostic counterpart, PoseFormer, and other state-of-the-art methods using up to hundreds of video frames regarding both speed and precision. Project page: https://qitaozhao.github.io/ContextAware-PoseFormer
Authors: Neng Dong, Shuanglin Yan, Hao Tang, Jinhui Tang, Liyan Zhang
Occluded person re-identification (re-ID) presents a challenging task due to occlusion perturbations. Although great efforts have been made to prevent the model from being disturbed by occlusion noise, most current solutions only capture information from a single image, disregarding the rich complementary information available in multiple images depicting the same pedestrian. In this paper, we propose a novel framework called Multi-view Information Integration and Propagation (MVI$^{2}$P). Specifically, realizing the potential of multi-view images in effectively characterizing the occluded target pedestrian, we integrate feature maps of which to create a comprehensive representation. During this process, to avoid introducing occlusion noise, we develop a CAMs-aware Localization module that selectively integrates information contributing to the identification. Additionally, considering the divergence in the discriminative nature of different images, we design a probability-aware Quantification module to emphatically integrate highly reliable information. Moreover, as multiple images with the same identity are not accessible in the testing stage, we devise an Information Propagation (IP) mechanism to distill knowledge from the comprehensive representation to that of a single occluded image. Extensive experiments and analyses have unequivocally demonstrated the effectiveness and superiority of the proposed MVI$^{2}$P. The code will be released at \url{https://github.com/nengdong96/MVIIP}.
Authors: Thi-Ngoc-Hanh Le, HuiGuang Huang, Yi-Ru Chen, Tong-Yee Lee
Video holds significance in computer graphics applications. Because of the heterogeneous of digital devices, retargeting videos becomes an essential function to enhance user viewing experience in such applications. In the research of video retargeting, preserving the relevant visual content in videos, avoiding flicking, and processing time are the vital challenges. Extending image retargeting techniques to the video domain is challenging due to the high running time. Prior work of video retargeting mainly utilizes time-consuming preprocessing to analyze frames. Plus, being tolerant of different video content, avoiding important objects from shrinking, and the ability to play with arbitrary ratios are the limitations that need to be resolved in these systems requiring investigation. In this paper, we present an end-to-end RETVI method to retarget videos to arbitrary aspect ratios. We eliminate the computational bottleneck in the conventional approaches by designing RETVI with two modules, content feature analyzer (CFA) and adaptive deforming estimator (ADE). The extensive experiments and evaluations show that our system outperforms previous work in quality and running time. Visit our project website for more results at this http URL
Authors: Yao Zhu, Yuefeng Chen, Wei Wang, Xiaofeng Mao, Xiu Yan, Yue Wang, Zhigang Li, Wang lu, Jindong Wang, Xiangyang Ji
Learning generalized representations from limited training samples is crucial for applying deep neural networks in low-resource scenarios. Recently, methods based on Contrastive Language-Image Pre-training (CLIP) have exhibited promising performance in few-shot adaptation tasks. To avoid catastrophic forgetting and overfitting caused by few-shot fine-tuning, existing works usually freeze the parameters of CLIP pre-trained on large-scale datasets, overlooking the possibility that some parameters might not be suitable for downstream tasks. To this end, we revisit CLIP's visual encoder with a specific focus on its distinctive attention pooling layer, which performs a spatial weighted-sum of the dense feature maps. Given that dense feature maps contain meaningful semantic information, and different semantics hold varying importance for diverse downstream tasks (such as prioritizing semantics like ears and eyes in pet classification tasks rather than side mirrors), using the same weighted-sum operation for dense features across different few-shot tasks might not be appropriate. Hence, we propose fine-tuning the parameters of the attention pooling layer during the training process to encourage the model to focus on task-specific semantics. In the inference process, we perform residual blending between the features pooled by the fine-tuned and the original attention pooling layers to incorporate both the few-shot knowledge and the pre-trained CLIP's prior knowledge. We term this method as Semantic-Aware FinE-tuning (SAFE). SAFE is effective in enhancing the conventional few-shot CLIP and is compatible with the existing adapter approach (termed SAFE-A).
Authors: Xiaoting Yin, Hao Shi, Jiaan Chen, Ze Wang, Yaozu Ye, Huajian Ni, Kailun Yang, Kaiwei Wang
Human pose estimation is a critical component in autonomous driving and parking, enhancing safety by predicting human actions. Traditional frame-based cameras and videos are commonly applied, yet, they become less reliable in scenarios under high dynamic range or heavy motion blur. In contrast, event cameras offer a robust solution for navigating these challenging contexts. Predominant methodologies incorporate event cameras into learning frameworks by accumulating events into event frames. However, such methods tend to marginalize the intrinsic asynchronous and high temporal resolution characteristics of events. This disregard leads to a loss in essential temporal dimension data, crucial for safety-critical tasks associated with dynamic human activities. To address this issue and to unlock the 3D potential of event information, we introduce two 3D event representations: the Rasterized Event Point Cloud (RasEPC) and the Decoupled Event Voxel (DEV). The RasEPC collates events within concise temporal slices at identical positions, preserving 3D attributes with statistical cues and markedly mitigating memory and computational demands. Meanwhile, the DEV representation discretizes events into voxels and projects them across three orthogonal planes, utilizing decoupled event attention to retrieve 3D cues from the 2D planes. Furthermore, we develop and release EV-3DPW, a synthetic event-based dataset crafted to facilitate training and quantitative analysis in outdoor scenes. On the public real-world DHP19 dataset, our event point cloud technique excels in real-time mobile predictions, while the decoupled event voxel method achieves the highest accuracy. Experiments reveal our proposed 3D representation methods' superior generalization capacities against traditional RGB images and event frame techniques. Our code and dataset are available at https://github.com/MasterHow/EventPointPose.
Authors: Suryaka Suresh, Bishshoy Das, Vinayak Abrol, Sumantra Dutta Roy
We study how the topology of feature embedding space changes as it passes through the layers of a well-trained deep neural network (DNN) through Betti numbers. Motivated by existing studies using simplicial complexes on shallow fully connected networks (FCN), we present an extended analysis using Cubical homology instead, with a variety of popular deep architectures and real image datasets. We demonstrate that as depth increases, a topologically complicated dataset is transformed into a simple one, resulting in Betti numbers attaining their lowest possible value. The rate of decay in topological complexity (as a metric) helps quantify the impact of architectural choices on the generalization ability. Interestingly from a representation learning perspective, we highlight several invariances such as topological invariance of (1) an architecture on similar datasets; (2) embedding space of a dataset for architectures of variable depth; (3) embedding space to input resolution/size, and (4) data sub-sampling. In order to further demonstrate the link between expressivity \& the generalization capability of a network, we consider the task of ranking pre-trained models for downstream classification task (transfer learning). Compared to existing approaches, the proposed metric has a better correlation to the actually achievable accuracy via fine-tuning the pre-trained model.
Authors: Rohit Venkata Sai Dulam, Chandra Kambhamettu
Developing a new Salient Object Detection (SOD) model involves selecting an ImageNet pre-trained backbone and creating novel feature refinement modules to use backbone features. However, adding new components to a pre-trained backbone needs retraining the whole network on the ImageNet dataset, which requires significant time. Hence, we explore developing a neural network from scratch directly trained on SOD without ImageNet pre-training. Such a formulation offers full autonomy to design task-specific components. To that end, we propose SODAWideNet, an encoder-decoder-style network for Salient Object Detection. We deviate from the commonly practiced paradigm of narrow and deep convolutional models to a wide and shallow architecture, resulting in a parameter-efficient deep neural network. To achieve a shallower network, we increase the receptive field from the beginning of the network using a combination of dilated convolutions and self-attention. Therefore, we propose Multi Receptive Field Feature Aggregation Module (MRFFAM) that efficiently obtains discriminative features from farther regions at higher resolutions using dilated convolutions. Next, we propose Multi-Scale Attention (MSA), which creates a feature pyramid and efficiently computes attention across multiple resolutions to extract global features from larger feature maps. Finally, we propose two variants, SODAWideNet-S (3.03M) and SODAWideNet (9.03M), that achieve competitive performance against state-of-the-art models on five datasets.