Pub Date : 2022-09-14DOI: 10.48550/arXiv.2209.06351
Kaichen Zhou, Lanqing Hong, Changhao Chen, Hang Xu, Chao Ye, Qingyong Hu, Zhenguo Li
Self-supervised depth learning from monocular images normally relies on the 2D pixel-wise photometric relation between temporally adjacent image frames. However, they neither fully exploit the 3D point-wise geometric correspondences, nor effectively tackle the ambiguities in the photometric warping caused by occlusions or illumination inconsistency. To address these problems, this work proposes Density Volume Construction Network (DevNet), a novel self-supervised monocular depth learning framework, that can consider 3D spatial information, and exploit stronger geometric constraints among adjacent camera frustums. Instead of directly regressing the pixel value from a single image, our DevNet divides the camera frustum into multiple parallel planes and predicts the pointwise occlusion probability density on each plane. The final depth map is generated by integrating the density along corresponding rays. During the training process, novel regularization strategies and loss functions are introduced to mitigate photometric ambiguities and overfitting. Without obviously enlarging model parameters size or running time, DevNet outperforms several representative baselines on both the KITTI-2015 outdoor dataset and NYU-V2 indoor dataset. In particular, the root-mean-square-deviation is reduced by around 4% with DevNet on both KITTI-2015 and NYU-V2 in the task of depth estimation. Code is available at https://github.com/gitkaichenzhou/DevNet.
{"title":"DevNet: Self-supervised Monocular Depth Learning via Density Volume Construction","authors":"Kaichen Zhou, Lanqing Hong, Changhao Chen, Hang Xu, Chao Ye, Qingyong Hu, Zhenguo Li","doi":"10.48550/arXiv.2209.06351","DOIUrl":"https://doi.org/10.48550/arXiv.2209.06351","url":null,"abstract":"Self-supervised depth learning from monocular images normally relies on the 2D pixel-wise photometric relation between temporally adjacent image frames. However, they neither fully exploit the 3D point-wise geometric correspondences, nor effectively tackle the ambiguities in the photometric warping caused by occlusions or illumination inconsistency. To address these problems, this work proposes Density Volume Construction Network (DevNet), a novel self-supervised monocular depth learning framework, that can consider 3D spatial information, and exploit stronger geometric constraints among adjacent camera frustums. Instead of directly regressing the pixel value from a single image, our DevNet divides the camera frustum into multiple parallel planes and predicts the pointwise occlusion probability density on each plane. The final depth map is generated by integrating the density along corresponding rays. During the training process, novel regularization strategies and loss functions are introduced to mitigate photometric ambiguities and overfitting. Without obviously enlarging model parameters size or running time, DevNet outperforms several representative baselines on both the KITTI-2015 outdoor dataset and NYU-V2 indoor dataset. In particular, the root-mean-square-deviation is reduced by around 4% with DevNet on both KITTI-2015 and NYU-V2 in the task of depth estimation. Code is available at https://github.com/gitkaichenzhou/DevNet.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"29 1","pages":"125-142"},"PeriodicalIF":0.0,"publicationDate":"2022-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85845112","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-12DOI: 10.48550/arXiv.2209.05624
Wufei Ma, Angtian Wang, A. Yuille, Adam Kortylewski
We consider the problem of category-level 6D pose estimation from a single RGB image. Our approach represents an object category as a cuboid mesh and learns a generative model of the neural feature activations at each mesh vertex to perform pose estimation through differentiable rendering. A common problem of rendering-based approaches is that they rely on bounding box proposals, which do not convey information about the 3D rotation of the object and are not reliable when objects are partially occluded. Instead, we introduce a coarse-to-fine optimization strategy that utilizes the rendering process to estimate a sparse set of 6D object proposals, which are subsequently refined with gradient-based optimization. The key to enabling the convergence of our approach is a neural feature representation that is trained to be scale- and rotation-invariant using contrastive learning. Our experiments demonstrate an enhanced category-level 6D pose estimation performance compared to prior work, particularly under strong partial occlusion.
{"title":"Robust Category-Level 6D Pose Estimation with Coarse-to-Fine Rendering of Neural Features","authors":"Wufei Ma, Angtian Wang, A. Yuille, Adam Kortylewski","doi":"10.48550/arXiv.2209.05624","DOIUrl":"https://doi.org/10.48550/arXiv.2209.05624","url":null,"abstract":"We consider the problem of category-level 6D pose estimation from a single RGB image. Our approach represents an object category as a cuboid mesh and learns a generative model of the neural feature activations at each mesh vertex to perform pose estimation through differentiable rendering. A common problem of rendering-based approaches is that they rely on bounding box proposals, which do not convey information about the 3D rotation of the object and are not reliable when objects are partially occluded. Instead, we introduce a coarse-to-fine optimization strategy that utilizes the rendering process to estimate a sparse set of 6D object proposals, which are subsequently refined with gradient-based optimization. The key to enabling the convergence of our approach is a neural feature representation that is trained to be scale- and rotation-invariant using contrastive learning. Our experiments demonstrate an enhanced category-level 6D pose estimation performance compared to prior work, particularly under strong partial occlusion.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"17 1","pages":"492-508"},"PeriodicalIF":0.0,"publicationDate":"2022-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77724814","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-12DOI: 10.48550/arXiv.2209.05588
Zixiang Zhou, Xian Zhao, Yu Wang, Panqu Wang, H. Foroosh
Query-based transformer has shown great potential in constructing long-range attention in many image-domain tasks, but has rarely been considered in LiDAR-based 3D object detection due to the overwhelming size of the point cloud data. In this paper, we propose CenterFormer, a center-based transformer network for 3D object detection. CenterFormer first uses a center heatmap to select center candidates on top of a standard voxel-based point cloud encoder. It then uses the feature of the center candidate as the query embedding in the transformer. To further aggregate features from multiple frames, we design an approach to fuse features through cross-attention. Lastly, regression heads are added to predict the bounding box on the output center feature representation. Our design reduces the convergence difficulty and computational complexity of the transformer structure. The results show significant improvements over the strong baseline of anchor-free object detection networks. CenterFormer achieves state-of-the-art performance for a single model on the Waymo Open Dataset, with 73.7% mAPH on the validation set and 75.6% mAPH on the test set, significantly outperforming all previously published CNN and transformer-based methods. Our code is publicly available at https://github.com/TuSimple/centerformer
{"title":"CenterFormer: Center-based Transformer for 3D Object Detection","authors":"Zixiang Zhou, Xian Zhao, Yu Wang, Panqu Wang, H. Foroosh","doi":"10.48550/arXiv.2209.05588","DOIUrl":"https://doi.org/10.48550/arXiv.2209.05588","url":null,"abstract":"Query-based transformer has shown great potential in constructing long-range attention in many image-domain tasks, but has rarely been considered in LiDAR-based 3D object detection due to the overwhelming size of the point cloud data. In this paper, we propose CenterFormer, a center-based transformer network for 3D object detection. CenterFormer first uses a center heatmap to select center candidates on top of a standard voxel-based point cloud encoder. It then uses the feature of the center candidate as the query embedding in the transformer. To further aggregate features from multiple frames, we design an approach to fuse features through cross-attention. Lastly, regression heads are added to predict the bounding box on the output center feature representation. Our design reduces the convergence difficulty and computational complexity of the transformer structure. The results show significant improvements over the strong baseline of anchor-free object detection networks. CenterFormer achieves state-of-the-art performance for a single model on the Waymo Open Dataset, with 73.7% mAPH on the validation set and 75.6% mAPH on the test set, significantly outperforming all previously published CNN and transformer-based methods. Our code is publicly available at https://github.com/TuSimple/centerformer","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"139 1","pages":"496-513"},"PeriodicalIF":0.0,"publicationDate":"2022-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79872637","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-08DOI: 10.48550/arXiv.2209.04027
Sk. Miraj Ahmed, Suhas Lohit, Kuan-Chuan Peng, Michael Jones, A. Roy-Chowdhury
Cost-effective depth and infrared sensors as alternatives to usual RGB sensors are now a reality, and have some advantages over RGB in domains like autonomous navigation and remote sensing. As such, building computer vision and deep learning systems for depth and infrared data are crucial. However, large labeled datasets for these modalities are still lacking. In such cases, transferring knowledge from a neural network trained on a well-labeled large dataset in the source modality (RGB) to a neural network that works on a target modality (depth, infrared, etc.) is of great value. For reasons like memory and privacy, it may not be possible to access the source data, and knowledge transfer needs to work with only the source models. We describe an effective solution, SOCKET: SOurce-free Cross-modal KnowledgE Transfer for this challenging task of transferring knowledge from one source modality to a different target modality without access to task-relevant source data. The framework reduces the modality gap using paired task-irrelevant data, as well as by matching the mean and variance of the target features with the batch-norm statistics that are present in the source models. We show through extensive experiments that our method significantly outperforms existing source-free methods for classification tasks which do not account for the modality gap.
{"title":"Cross-Modal Knowledge Transfer Without Task-Relevant Source Data","authors":"Sk. Miraj Ahmed, Suhas Lohit, Kuan-Chuan Peng, Michael Jones, A. Roy-Chowdhury","doi":"10.48550/arXiv.2209.04027","DOIUrl":"https://doi.org/10.48550/arXiv.2209.04027","url":null,"abstract":"Cost-effective depth and infrared sensors as alternatives to usual RGB sensors are now a reality, and have some advantages over RGB in domains like autonomous navigation and remote sensing. As such, building computer vision and deep learning systems for depth and infrared data are crucial. However, large labeled datasets for these modalities are still lacking. In such cases, transferring knowledge from a neural network trained on a well-labeled large dataset in the source modality (RGB) to a neural network that works on a target modality (depth, infrared, etc.) is of great value. For reasons like memory and privacy, it may not be possible to access the source data, and knowledge transfer needs to work with only the source models. We describe an effective solution, SOCKET: SOurce-free Cross-modal KnowledgE Transfer for this challenging task of transferring knowledge from one source modality to a different target modality without access to task-relevant source data. The framework reduces the modality gap using paired task-irrelevant data, as well as by matching the mean and variance of the target features with the batch-norm statistics that are present in the source models. We show through extensive experiments that our method significantly outperforms existing source-free methods for classification tasks which do not account for the modality gap.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"7 1","pages":"111-127"},"PeriodicalIF":0.0,"publicationDate":"2022-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88568569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-08DOI: 10.48550/arXiv.2209.03592
P. Wang, Cheng Da, C. Yao
. Scene text recognition (STR) has been an active research topic in computer vision for years. To tackle this challenging problem, numerous innovative methods have been successively proposed and incorporating linguistic knowledge into STR models has recently become a prominent trend. In this work, we first draw inspiration from the recent progress in Vision Transformer (ViT) to construct a conceptually simple yet powerful vision STR model, which is built upon ViT and outperforms previous state-of-the-art models for scene text recognition, including both pure vision models and language-augmented methods. To integrate linguistic knowledge, we further propose a Multi-Granularity Prediction strategy to inject information from the language modality into the model in an implicit way, i.e. , subword representations (BPE and WordPiece) widely-used in NLP are introduced into the output space, in addition to the conventional character level representation, while no independent language model (LM) is adopted. The resultant algorithm (termed MGP-STR) is able to push the performance envelop of STR to an even higher level. Specifically, it achieves an average recognition accuracy of 93 . 35% on standard benchmarks. Code will be released soon.
{"title":"Multi-Granularity Prediction for Scene Text Recognition","authors":"P. Wang, Cheng Da, C. Yao","doi":"10.48550/arXiv.2209.03592","DOIUrl":"https://doi.org/10.48550/arXiv.2209.03592","url":null,"abstract":". Scene text recognition (STR) has been an active research topic in computer vision for years. To tackle this challenging problem, numerous innovative methods have been successively proposed and incorporating linguistic knowledge into STR models has recently become a prominent trend. In this work, we first draw inspiration from the recent progress in Vision Transformer (ViT) to construct a conceptually simple yet powerful vision STR model, which is built upon ViT and outperforms previous state-of-the-art models for scene text recognition, including both pure vision models and language-augmented methods. To integrate linguistic knowledge, we further propose a Multi-Granularity Prediction strategy to inject information from the language modality into the model in an implicit way, i.e. , subword representations (BPE and WordPiece) widely-used in NLP are introduced into the output space, in addition to the conventional character level representation, while no independent language model (LM) is adopted. The resultant algorithm (termed MGP-STR) is able to push the performance envelop of STR to an even higher level. Specifically, it achieves an average recognition accuracy of 93 . 35% on standard benchmarks. Code will be released soon.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"112 1","pages":"339-355"},"PeriodicalIF":0.0,"publicationDate":"2022-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80658762","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-06DOI: 10.48550/arXiv.2209.02686
Justin D. Theiss, Jay Leverett, Daeil Kim, Aayush Prakash
Image-to-image translation has played an important role in enabling synthetic data for computer vision. However, if the source and target domains have a large semantic mismatch, existing techniques often suffer from source content corruption aka semantic flipping. To address this problem, we propose a new paradigm for image-to-image translation using Vector Symbolic Architectures (VSA), a theoretical framework which defines algebraic operations in a high-dimensional vector (hypervector) space. We introduce VSA-based constraints on adversarial learning for source-to-target translations by learning a hypervector mapping that inverts the translation to ensure consistency with source content. We show both qualitatively and quantitatively that our method improves over other state-of-the-art techniques.
{"title":"Unpaired Image Translation via Vector Symbolic Architectures","authors":"Justin D. Theiss, Jay Leverett, Daeil Kim, Aayush Prakash","doi":"10.48550/arXiv.2209.02686","DOIUrl":"https://doi.org/10.48550/arXiv.2209.02686","url":null,"abstract":"Image-to-image translation has played an important role in enabling synthetic data for computer vision. However, if the source and target domains have a large semantic mismatch, existing techniques often suffer from source content corruption aka semantic flipping. To address this problem, we propose a new paradigm for image-to-image translation using Vector Symbolic Architectures (VSA), a theoretical framework which defines algebraic operations in a high-dimensional vector (hypervector) space. We introduce VSA-based constraints on adversarial learning for source-to-target translations by learning a hypervector mapping that inverts the translation to ensure consistency with source content. We show both qualitatively and quantitatively that our method improves over other state-of-the-art techniques.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"8 1","pages":"17-32"},"PeriodicalIF":0.0,"publicationDate":"2022-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75183219","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
, Abstract. Existing Binary Neural Networks (BNNs) mainly operate on local convolutions with binarization function. However, such simple bit operations lack the ability of modeling contextual dependencies, which is critical for learning discriminative deep representations in vision models. In this work, we tackle this issue by presenting new designs of binary neural modules, which enables BNNs to learn effective contextual dependencies. First, we propose a binary multi-layer perceptron (MLP) block as an alternative to binary convolution blocks to directly model contextual dependencies. Both short-range and long-range feature dependencies are modeled by binary MLPs, where the former provides local inductive bias and the latter breaks limited receptive field in binary convolutions. Second, to improve the robustness of binary models with contextual dependencies, we compute the contextual dynamic embeddings to determine the binarization thresholds in general binary convolutional blocks. Armed with our binary MLP blocks and improved binary convolution, we build the BNNs with explicit Contextual De-pendency modeling, termed as BCDNet. On the standard ImageNet-1K classification benchmark, the BCDNet achieves 72.3% Top-1 accuracy and outperforms leading binary methods by a large margin. In particu-lar, the proposed BCDNet exceeds the state-of-the-art ReActNet-A by 2.9% Top-1 accuracy with similar operations. Our code is available at https://github.com/Sense-GVT/BCDNet .
{"title":"Towards Accurate Binary Neural Networks via Modeling Contextual Dependencies","authors":"Xingrun Xing, Yangguang Li, Wei Li, Wenrui Ding, Yalong Jiang, Yufeng Wang, Jinghua Shao, Chunlei Liu, Xianglong Liu","doi":"10.48550/arXiv.2209.01404","DOIUrl":"https://doi.org/10.48550/arXiv.2209.01404","url":null,"abstract":", Abstract. Existing Binary Neural Networks (BNNs) mainly operate on local convolutions with binarization function. However, such simple bit operations lack the ability of modeling contextual dependencies, which is critical for learning discriminative deep representations in vision models. In this work, we tackle this issue by presenting new designs of binary neural modules, which enables BNNs to learn effective contextual dependencies. First, we propose a binary multi-layer perceptron (MLP) block as an alternative to binary convolution blocks to directly model contextual dependencies. Both short-range and long-range feature dependencies are modeled by binary MLPs, where the former provides local inductive bias and the latter breaks limited receptive field in binary convolutions. Second, to improve the robustness of binary models with contextual dependencies, we compute the contextual dynamic embeddings to determine the binarization thresholds in general binary convolutional blocks. Armed with our binary MLP blocks and improved binary convolution, we build the BNNs with explicit Contextual De-pendency modeling, termed as BCDNet. On the standard ImageNet-1K classification benchmark, the BCDNet achieves 72.3% Top-1 accuracy and outperforms leading binary methods by a large margin. In particu-lar, the proposed BCDNet exceeds the state-of-the-art ReActNet-A by 2.9% Top-1 accuracy with similar operations. Our code is available at https://github.com/Sense-GVT/BCDNet .","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"8 1","pages":"536-552"},"PeriodicalIF":0.0,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86602070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-03DOI: 10.48550/arXiv.2209.01501
Zhenyi Wang, Li Shen, Le Fang, Qiuling Suo, Dongling Zhan, Tiehang Duan, Mingchen Gao
. The paradigm of machine intelligence moves from purely supervised learning to a more practical scenario when many loosely related unlabeled data are available and labeled data is scarce. Most existing algo-rithms assume that the underlying task distribution is stationary. Here we consider a more realistic and challenging setting in that task distributions evolve over time. We name this problem as S emi-supervised meta-learning with E volving T ask di S tributions, abbreviated as SETS . Two key challenges arise in this more realistic setting: (i) how to use unlabeled data in the presence of a large amount of unlabeled out-of-distribution (OOD) data; and (ii) how to prevent catastrophic forgetting on previously learned task distributions due to the task distribution shift. We propose an O OD R obust and knowle D ge pres E rved semi-supe R vised meta-learning approach ( ORDER ) ‡ , to tackle these two major challenges. Specifically, our ORDER introduces a novel mutual information regularization to robustify the model with unlabeled OOD data and adopts an optimal transport regularization to remember previously learned knowledge in feature space. In addition, we test our method on a very challenging dataset: SETS on large-scale non-stationary semi-supervised task distributions consisting of (at least) 72K tasks. With extensive experiments, we demonstrate the proposed ORDER alleviates forgetting on evolving task distributions and is more robust to OOD data than related strong baselines.
{"title":"Meta-Learning with Less Forgetting on Large-Scale Non-Stationary Task Distributions","authors":"Zhenyi Wang, Li Shen, Le Fang, Qiuling Suo, Dongling Zhan, Tiehang Duan, Mingchen Gao","doi":"10.48550/arXiv.2209.01501","DOIUrl":"https://doi.org/10.48550/arXiv.2209.01501","url":null,"abstract":". The paradigm of machine intelligence moves from purely supervised learning to a more practical scenario when many loosely related unlabeled data are available and labeled data is scarce. Most existing algo-rithms assume that the underlying task distribution is stationary. Here we consider a more realistic and challenging setting in that task distributions evolve over time. We name this problem as S emi-supervised meta-learning with E volving T ask di S tributions, abbreviated as SETS . Two key challenges arise in this more realistic setting: (i) how to use unlabeled data in the presence of a large amount of unlabeled out-of-distribution (OOD) data; and (ii) how to prevent catastrophic forgetting on previously learned task distributions due to the task distribution shift. We propose an O OD R obust and knowle D ge pres E rved semi-supe R vised meta-learning approach ( ORDER ) ‡ , to tackle these two major challenges. Specifically, our ORDER introduces a novel mutual information regularization to robustify the model with unlabeled OOD data and adopts an optimal transport regularization to remember previously learned knowledge in feature space. In addition, we test our method on a very challenging dataset: SETS on large-scale non-stationary semi-supervised task distributions consisting of (at least) 72K tasks. With extensive experiments, we demonstrate the proposed ORDER alleviates forgetting on evolving task distributions and is more robust to OOD data than related strong baselines.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"38 1","pages":"221-238"},"PeriodicalIF":0.0,"publicationDate":"2022-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86359811","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-02DOI: 10.48550/arXiv.2209.01199
Ali Dabouei, Fariborz Taherkhani, Sobhan Soleymani, N. Nasrabadi
. Despite the fundamental distinction between adversarial and natural training (AT and NT), AT methods generally adopt momentum SGD (MSGD) for the outer optimization. This paper aims to analyze this choice by investigating the overlooked role of outer optimization in AT. Our exploratory evaluations reveal that AT induces higher gradient norm and variance compared to NT. This phenomenon hinders the outer optimization in AT since the convergence rate of MSGD is highly dependent on the variance of the gradients. To this end, we propose an optimization method called ENGM which regularizes the contribution of each input example to the average mini-batch gradients. We prove that the convergence rate of ENGM is independent of the variance of the gradients, and thus, it is suitable for AT. We introduce a trick to reduce the computational cost of ENGM using empirical observations on the correlation between the norm of gradients w.r.t. the network parameters and input examples. Our extensive evaluations and ablation studies on CIFAR-10, CIFAR-100, and TinyImageNet demonstrate that ENGM and its variants consistently improve the performance of a wide range of AT methods. Furthermore, ENGM alleviates major shortcomings of AT including robust overfitting and high sensitivity to hyperparameter settings.
{"title":"Revisiting Outer Optimization in Adversarial Training","authors":"Ali Dabouei, Fariborz Taherkhani, Sobhan Soleymani, N. Nasrabadi","doi":"10.48550/arXiv.2209.01199","DOIUrl":"https://doi.org/10.48550/arXiv.2209.01199","url":null,"abstract":". Despite the fundamental distinction between adversarial and natural training (AT and NT), AT methods generally adopt momentum SGD (MSGD) for the outer optimization. This paper aims to analyze this choice by investigating the overlooked role of outer optimization in AT. Our exploratory evaluations reveal that AT induces higher gradient norm and variance compared to NT. This phenomenon hinders the outer optimization in AT since the convergence rate of MSGD is highly dependent on the variance of the gradients. To this end, we propose an optimization method called ENGM which regularizes the contribution of each input example to the average mini-batch gradients. We prove that the convergence rate of ENGM is independent of the variance of the gradients, and thus, it is suitable for AT. We introduce a trick to reduce the computational cost of ENGM using empirical observations on the correlation between the norm of gradients w.r.t. the network parameters and input examples. Our extensive evaluations and ablation studies on CIFAR-10, CIFAR-100, and TinyImageNet demonstrate that ENGM and its variants consistently improve the performance of a wide range of AT methods. Furthermore, ENGM alleviates major shortcomings of AT including robust overfitting and high sensitivity to hyperparameter settings.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"1 1","pages":"244-261"},"PeriodicalIF":0.0,"publicationDate":"2022-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88710069","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-01DOI: 10.48550/arXiv.2209.00638
Nadine Behrmann, S. Golestaneh, Zico Kolter, Juergen Gall, M. Noroozi
This paper introduces a unified framework for video action segmentation via sequence to sequence (seq2seq) translation in a fully and timestamp supervised setup. In contrast to current state-of-the-art frame-level prediction methods, we view action segmentation as a seq2seq translation task, i.e., mapping a sequence of video frames to a sequence of action segments. Our proposed method involves a series of modifications and auxiliary loss functions on the standard Transformer seq2seq translation model to cope with long input sequences opposed to short output sequences and relatively few videos. We incorporate an auxiliary supervision signal for the encoder via a frame-wise loss and propose a separate alignment decoder for an implicit duration prediction. Finally, we extend our framework to the timestamp supervised setting via our proposed constrained k-medoids algorithm to generate pseudo-segmentations. Our proposed framework performs consistently on both fully and timestamp supervised settings, outperforming or competing state-of-the-art on several datasets. Our code is publicly available at https://github.com/boschresearch/UVAST.
{"title":"Unified Fully and Timestamp Supervised Temporal Action Segmentation via Sequence to Sequence Translation","authors":"Nadine Behrmann, S. Golestaneh, Zico Kolter, Juergen Gall, M. Noroozi","doi":"10.48550/arXiv.2209.00638","DOIUrl":"https://doi.org/10.48550/arXiv.2209.00638","url":null,"abstract":"This paper introduces a unified framework for video action segmentation via sequence to sequence (seq2seq) translation in a fully and timestamp supervised setup. In contrast to current state-of-the-art frame-level prediction methods, we view action segmentation as a seq2seq translation task, i.e., mapping a sequence of video frames to a sequence of action segments. Our proposed method involves a series of modifications and auxiliary loss functions on the standard Transformer seq2seq translation model to cope with long input sequences opposed to short output sequences and relatively few videos. We incorporate an auxiliary supervision signal for the encoder via a frame-wise loss and propose a separate alignment decoder for an implicit duration prediction. Finally, we extend our framework to the timestamp supervised setting via our proposed constrained k-medoids algorithm to generate pseudo-segmentations. Our proposed framework performs consistently on both fully and timestamp supervised settings, outperforming or competing state-of-the-art on several datasets. Our code is publicly available at https://github.com/boschresearch/UVAST.","PeriodicalId":72676,"journal":{"name":"Computer vision - ECCV ... : ... European Conference on Computer Vision : proceedings. European Conference on Computer Vision","volume":"5 1","pages":"52-68"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83706870","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：