Pub Date : 2021-10-01DOI: 10.1109/ICCV48922.2021.00506
Jae-Han Lee, Chulwoo Lee, Chang-Su Kim
We propose a novel loss weighting algorithm, called loss scale balancing (LSB), for multi-task learning (MTL) of pixelwise vision tasks. An MTL model is trained to estimate multiple pixelwise predictions using an overall loss, which is a linear combination of individual task losses. The proposed algorithm dynamically adjusts the linear weights to learn all tasks effectively. Instead of controlling the trend of each loss value directly, we balance the loss scale — the product of the loss value and its weight — periodically. In addition, by evaluating the difficulty of each task based on the previous loss record, the proposed algorithm focuses more on difficult tasks during training. Experimental results show that the proposed algorithm outperforms conventional weighting algorithms for MTL of various pixelwise tasks. Codes are available at https://github.com/jaehanlee-mcl/LSB-MTL.
{"title":"Learning Multiple Pixelwise Tasks Based on Loss Scale Balancing","authors":"Jae-Han Lee, Chulwoo Lee, Chang-Su Kim","doi":"10.1109/ICCV48922.2021.00506","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00506","url":null,"abstract":"We propose a novel loss weighting algorithm, called loss scale balancing (LSB), for multi-task learning (MTL) of pixelwise vision tasks. An MTL model is trained to estimate multiple pixelwise predictions using an overall loss, which is a linear combination of individual task losses. The proposed algorithm dynamically adjusts the linear weights to learn all tasks effectively. Instead of controlling the trend of each loss value directly, we balance the loss scale — the product of the loss value and its weight — periodically. In addition, by evaluating the difficulty of each task based on the previous loss record, the proposed algorithm focuses more on difficult tasks during training. Experimental results show that the proposed algorithm outperforms conventional weighting algorithms for MTL of various pixelwise tasks. Codes are available at https://github.com/jaehanlee-mcl/LSB-MTL.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"32 1","pages":"5087-5096"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75070997","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00173
Nayoung Kim, S. Ha, Jewon Kang
Video Question Answering (Video QA) aims to give an answer to the question through semantic reasoning between visual and linguistic information. Recently, handling large amounts of multi-modal video and language information of a video is considered important in the industry. However, the current video QA models use deep features, suffered from significant computational complexity and insufficient representation capability both in training and testing. Existing features are extracted using pre-trained networks after all the frames are decoded, which is not always suitable for video QA tasks. In this paper, we develop a novel deep neural network to provide video QA features obtained from coded video bit-stream to reduce the complexity. The proposed network includes several dedicated deep modules to both the video QA and the video compression system, which is the first attempt at the video QA task. The proposed network is predominantly model-agnostic. It is integrated into the state-of-the-art networks for improved performance without any computationally expensive motion-related deep models. The experimental results demonstrate that the proposed network outperforms the previous studies at lower complexity. https://github.com/Nayoung-Kim-ICP/VQAC
视频问答(Video Question answer, Video QA)旨在通过视觉信息和语言信息之间的语义推理，给出问题的答案。近年来，处理大量的多模态视频和视频的语言信息被业界认为是很重要的。然而，目前的视频QA模型使用深度特征，在训练和测试中都存在计算复杂度和表示能力不足的问题。在所有帧解码后，使用预训练的网络提取现有特征，这并不总是适用于视频QA任务。在本文中，我们开发了一种新的深度神经网络来提供从编码视频比特流中获得的视频QA特征，以降低复杂度。该网络包含了多个专用于视频质量保证和视频压缩系统的深度模块，是视频质量保证任务的首次尝试。所提出的网络主要是模型不可知的。它被集成到最先进的网络中，以提高性能，而无需任何计算昂贵的运动相关深度模型。实验结果表明，该网络在较低的复杂度下优于前人的研究。https://github.com/Nayoung-Kim-ICP/VQAC
{"title":"Video Question Answering Using Language-Guided Deep Compressed-Domain Video Feature","authors":"Nayoung Kim, S. Ha, Jewon Kang","doi":"10.1109/ICCV48922.2021.00173","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00173","url":null,"abstract":"Video Question Answering (Video QA) aims to give an answer to the question through semantic reasoning between visual and linguistic information. Recently, handling large amounts of multi-modal video and language information of a video is considered important in the industry. However, the current video QA models use deep features, suffered from significant computational complexity and insufficient representation capability both in training and testing. Existing features are extracted using pre-trained networks after all the frames are decoded, which is not always suitable for video QA tasks. In this paper, we develop a novel deep neural network to provide video QA features obtained from coded video bit-stream to reduce the complexity. The proposed network includes several dedicated deep modules to both the video QA and the video compression system, which is the first attempt at the video QA task. The proposed network is predominantly model-agnostic. It is integrated into the state-of-the-art networks for improved performance without any computationally expensive motion-related deep models. The experimental results demonstrate that the proposed network outperforms the previous studies at lower complexity. https://github.com/Nayoung-Kim-ICP/VQAC","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"44 1","pages":"1688-1697"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74948769","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00057
Shuyang Sun, Xiaoyu Yue, Xiaojuan Qi, Wanli Ouyang, V. Prisacariu, Philip H. S. Torr
Aggregating features from different depths of a network is widely adopted to improve the network capability. Lots of modern architectures are equipped with skip connections, which actually makes the feature aggregation happen in all these networks. Since different features tell different semantic meanings, there are inconsistencies and incompatibilities to be solved. However, existing works naively blend deep features via element-wise summation or concatenation with a convolution behind. Better feature aggregation method beyond summation or concatenation is rarely explored. In this paper, given two layers of features to be aggregated together, we first detect and identify where and what needs to be updated in one layer, then replace the feature at the identified location with the information of the other layer This process, which we call DEtect-rePLAce (DEPLA), enables us to avoid inconsistent patterns while keeping useful information in the merged outputs. Experimental results demonstrate our method largely boosts multiple baselines e.g. ResNet, FishNet and FPN on three major vision tasks including ImageNet classification, MS COCO object detection and instance segmentation.
{"title":"Aggregation with Feature Detection","authors":"Shuyang Sun, Xiaoyu Yue, Xiaojuan Qi, Wanli Ouyang, V. Prisacariu, Philip H. S. Torr","doi":"10.1109/ICCV48922.2021.00057","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00057","url":null,"abstract":"Aggregating features from different depths of a network is widely adopted to improve the network capability. Lots of modern architectures are equipped with skip connections, which actually makes the feature aggregation happen in all these networks. Since different features tell different semantic meanings, there are inconsistencies and incompatibilities to be solved. However, existing works naively blend deep features via element-wise summation or concatenation with a convolution behind. Better feature aggregation method beyond summation or concatenation is rarely explored. In this paper, given two layers of features to be aggregated together, we first detect and identify where and what needs to be updated in one layer, then replace the feature at the identified location with the information of the other layer This process, which we call DEtect-rePLAce (DEPLA), enables us to avoid inconsistent patterns while keeping useful information in the merged outputs. Experimental results demonstrate our method largely boosts multiple baselines e.g. ResNet, FishNet and FPN on three major vision tasks including ImageNet classification, MS COCO object detection and instance segmentation.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"67 1","pages":"507-516"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74986947","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00414
Xiaotian Qiao, G. Hancke, Rynson W. H. Lau
Causally-taken images often suffer from flare artifacts, due to the unintended reflections and scattering of light inside the camera. However, as flares may appear in a variety of shapes, positions, and colors, detecting and removing them entirely from an image is very challenging. Existing methods rely on predefined intensity and geometry priors of flares, and may fail to distinguish the difference between light sources and flare artifacts. We observe that the conditions of the light source in the image play an important role in the resulting flares. In this paper, we present a deep framework with light source aware guidance for single-image flare removal (SIFR). In particular, we first detect the light source regions and the flare regions separately, and then remove the flare artifacts based on the light source aware guidance. By learning the underlying relationships between the two types of regions, our approach can remove different kinds of flares from the image. In addition, instead of using paired training data which are difficult to collect, we propose the first unpaired flare removal dataset and new cycle-consistency constraints to obtain more diverse examples and avoid manual annotations. Extensive experiments demonstrate that our method outperforms the baselines qualitatively and quantitatively. We also show that our model can be applied to flare effect manipulation (e.g., adding or changing image flares).
{"title":"Light Source Guided Single-Image Flare Removal from Unpaired Data","authors":"Xiaotian Qiao, G. Hancke, Rynson W. H. Lau","doi":"10.1109/ICCV48922.2021.00414","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00414","url":null,"abstract":"Causally-taken images often suffer from flare artifacts, due to the unintended reflections and scattering of light inside the camera. However, as flares may appear in a variety of shapes, positions, and colors, detecting and removing them entirely from an image is very challenging. Existing methods rely on predefined intensity and geometry priors of flares, and may fail to distinguish the difference between light sources and flare artifacts. We observe that the conditions of the light source in the image play an important role in the resulting flares. In this paper, we present a deep framework with light source aware guidance for single-image flare removal (SIFR). In particular, we first detect the light source regions and the flare regions separately, and then remove the flare artifacts based on the light source aware guidance. By learning the underlying relationships between the two types of regions, our approach can remove different kinds of flares from the image. In addition, instead of using paired training data which are difficult to collect, we propose the first unpaired flare removal dataset and new cycle-consistency constraints to obtain more diverse examples and avoid manual annotations. Extensive experiments demonstrate that our method outperforms the baselines qualitatively and quantitatively. We also show that our model can be applied to flare effect manipulation (e.g., adding or changing image flares).","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"1 1","pages":"4157-4165"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72958918","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}
Image-level weakly supervised semantic segmentation is a challenging task. As classification networks tend to capture notable object features and are insensitive to over-activation, class activation map (CAM) is too sparse and rough to guide segmentation network training. Inspired by the fact that erasing distinguishing features force networks to collect new ones from non-discriminative object regions, we using relationships between CAMs to propose a novel weakly supervised method. In this work, we apply these features, learned from erased images, as segmentation super-vision, driving network to study robust representation. In specifically, object regions obtained by CAM techniques are erased on images firstly. To provide other regions with seg-mentation supervision, Erased CAM Supervision Net (ECS-Net) generates pixel-level labels by predicting segmentation results of those processed images. We also design the rule of suppressing noise to select reliable labels. Our experiments on PASCAL VOC 2012 dataset show that without data annotations except for ground truth image-level labels, our ECS-Net achieves 67.6% mIoU on test set and 66.6% mIoU on val set, outperforming previous state-of-the-art methods.
{"title":"ECS-Net: Improving Weakly Supervised Semantic Segmentation by Using Connections Between Class Activation Maps","authors":"Kunyang Sun, Haoqing Shi, Zhengming Zhang, Yongming Huang","doi":"10.1109/ICCV48922.2021.00719","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00719","url":null,"abstract":"Image-level weakly supervised semantic segmentation is a challenging task. As classification networks tend to capture notable object features and are insensitive to over-activation, class activation map (CAM) is too sparse and rough to guide segmentation network training. Inspired by the fact that erasing distinguishing features force networks to collect new ones from non-discriminative object regions, we using relationships between CAMs to propose a novel weakly supervised method. In this work, we apply these features, learned from erased images, as segmentation super-vision, driving network to study robust representation. In specifically, object regions obtained by CAM techniques are erased on images firstly. To provide other regions with seg-mentation supervision, Erased CAM Supervision Net (ECS-Net) generates pixel-level labels by predicting segmentation results of those processed images. We also design the rule of suppressing noise to select reliable labels. Our experiments on PASCAL VOC 2012 dataset show that without data annotations except for ground truth image-level labels, our ECS-Net achieves 67.6% mIoU on test set and 66.6% mIoU on val set, outperforming previous state-of-the-art methods.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"13 1","pages":"7263-7272"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73007672","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.01117
Linlin Yang, Shicheng Chen, Angela Yao
We present SemiHand, a semi-supervised framework for 3D hand pose estimation from monocular images. We pre-train the model on labelled synthetic data and fine-tune it on unlabelled real-world data by pseudo-labeling with consistency training. By design, we introduce data augmentation of differing difficulties, consistency regularizer, label correction and sample selection for RGB-based 3D hand pose estimation. In particular, by approximating the hand masks from hand poses, we propose a cross-modal consistency and leverage semantic predictions to guide the predicted poses. Meanwhile, we introduce pose registration as label correction to guarantee the biomechanical feasibility of hand bone lengths. Experiments show that our method achieves a favorable improvement on real-world datasets after fine-tuning.
{"title":"SemiHand: Semi-supervised Hand Pose Estimation with Consistency","authors":"Linlin Yang, Shicheng Chen, Angela Yao","doi":"10.1109/ICCV48922.2021.01117","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.01117","url":null,"abstract":"We present SemiHand, a semi-supervised framework for 3D hand pose estimation from monocular images. We pre-train the model on labelled synthetic data and fine-tune it on unlabelled real-world data by pseudo-labeling with consistency training. By design, we introduce data augmentation of differing difficulties, consistency regularizer, label correction and sample selection for RGB-based 3D hand pose estimation. In particular, by approximating the hand masks from hand poses, we propose a cross-modal consistency and leverage semantic predictions to guide the predicted poses. Meanwhile, we introduce pose registration as label correction to guarantee the biomechanical feasibility of hand bone lengths. Experiments show that our method achieves a favorable improvement on real-world datasets after fine-tuning.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"15 1","pages":"11344-11353"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73281822","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00499
Yuqi Ding, Yu Ji, Mingyuan Zhou, S. B. Kang, Jinwei Ye
Helmholtz stereopsis (HS) exploits the reciprocity principle of light propagation (i.e., the Helmholtz reciprocity) for 3D reconstruction of surfaces with arbitrary reflectance. In this paper, we present the polarimetric Helmholtz stereopsis (polar-HS), which extends the classical HS by considering the polarization state of light in the reciprocal paths. With the additional phase information from polarization, polar-HS requires only one reciprocal image pair. We formulate new reciprocity and diffuse/specular polarimetric constraints to recover surface depths and normals using an optimization framework. Using a hardware prototype, we show that our approach produces high-quality 3D reconstruction for different types of surfaces, ranging from diffuse to highly specular.
{"title":"Polarimetric Helmholtz Stereopsis","authors":"Yuqi Ding, Yu Ji, Mingyuan Zhou, S. B. Kang, Jinwei Ye","doi":"10.1109/ICCV48922.2021.00499","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00499","url":null,"abstract":"Helmholtz stereopsis (HS) exploits the reciprocity principle of light propagation (i.e., the Helmholtz reciprocity) for 3D reconstruction of surfaces with arbitrary reflectance. In this paper, we present the polarimetric Helmholtz stereopsis (polar-HS), which extends the classical HS by considering the polarization state of light in the reciprocal paths. With the additional phase information from polarization, polar-HS requires only one reciprocal image pair. We formulate new reciprocity and diffuse/specular polarimetric constraints to recover surface depths and normals using an optimization framework. Using a hardware prototype, we show that our approach produces high-quality 3D reconstruction for different types of surfaces, ranging from diffuse to highly specular.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"23 3","pages":"5017-5026"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"72586627","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00158
Miaohui Zhang, Jie Liu, Yifei Wang, Yongri Piao, S. Yao, Wei Ji, Jingjing Li, Huchuan Lu, Zhongxuan Luo
The ability to capture inter-frame dynamics has been critical to the development of video salient object detection (VSOD). While many works have achieved great success in this field, a deeper insight into its dynamic nature should be developed. In this work, we aim to answer the following questions: How can a model adjust itself to dynamic variations as well as perceive fine differences in the real-world environment; How are the temporal dynamics well introduced into spatial information over time? To this end, we propose a dynamic context-sensitive filtering network (DCFNet) equipped with a dynamic context-sensitive filtering module (DCFM) and an effective bidirectional dynamic fusion strategy. The proposed DCFM sheds new light on dynamic filter generation by extracting location-related affinities between consecutive frames. Our bidirectional dynamic fusion strategy encourages the interaction of spatial and temporal information in a dynamic manner. Experimental results demonstrate that our proposed method can achieve state-of-the-art performance on most VSOD datasets while ensuring a real-time speed of 28 fps. The source code is publicly available at https://github.com/OIPLab-DUT/DCFNet.
{"title":"Dynamic Context-Sensitive Filtering Network for Video Salient Object Detection","authors":"Miaohui Zhang, Jie Liu, Yifei Wang, Yongri Piao, S. Yao, Wei Ji, Jingjing Li, Huchuan Lu, Zhongxuan Luo","doi":"10.1109/ICCV48922.2021.00158","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00158","url":null,"abstract":"The ability to capture inter-frame dynamics has been critical to the development of video salient object detection (VSOD). While many works have achieved great success in this field, a deeper insight into its dynamic nature should be developed. In this work, we aim to answer the following questions: How can a model adjust itself to dynamic variations as well as perceive fine differences in the real-world environment; How are the temporal dynamics well introduced into spatial information over time? To this end, we propose a dynamic context-sensitive filtering network (DCFNet) equipped with a dynamic context-sensitive filtering module (DCFM) and an effective bidirectional dynamic fusion strategy. The proposed DCFM sheds new light on dynamic filter generation by extracting location-related affinities between consecutive frames. Our bidirectional dynamic fusion strategy encourages the interaction of spatial and temporal information in a dynamic manner. Experimental results demonstrate that our proposed method can achieve state-of-the-art performance on most VSOD datasets while ensuring a real-time speed of 28 fps. The source code is publicly available at https://github.com/OIPLab-DUT/DCFNet.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"59 1","pages":"1533-1543"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73677896","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00038
Kun Yuan, Quanquan Li, Shaopeng Guo, Dapeng Chen, Aojun Zhou, F. Yu, Ziwei Liu
A standard practice of deploying deep neural networks is to apply the same architecture to all the input instances. However, a fixed architecture may not be suitable for different data with high diversity. To boost the model capacity, existing methods usually employ larger convolutional kernels or deeper network layers, which incurs prohibitive computational costs. In this paper, we address this issue by proposing Differentiable Dynamic Wirings (DDW), which learns the instance-aware connectivity that creates different wiring patterns for different instances. 1) Specifically, the network is initialized as a complete directed acyclic graph, where the nodes represent convolutional blocks and the edges represent the connection paths. 2) We generate edge weights by a learnable module, Router, and select the edges whose weights are larger than a threshold, to adjust the connectivity of the neural network structure. 3) Instead of using the same path of the network, DDW aggregates features dynamically in each node, which allows the network to have more representation power.To facilitate effective training, we further represent the network connectivity of each sample as an adjacency matrix. The matrix is updated to aggregate features in the forward pass, cached in the memory, and used for gradient computing in the backward pass. We validate the effectiveness of our approach with several mainstream architectures, including MobileNetV2, ResNet, ResNeXt, and RegNet. Extensive experiments are performed on ImageNet classification and COCO object detection, which demonstrates the effectiveness and generalization ability of our approach.
{"title":"Differentiable Dynamic Wirings for Neural Networks","authors":"Kun Yuan, Quanquan Li, Shaopeng Guo, Dapeng Chen, Aojun Zhou, F. Yu, Ziwei Liu","doi":"10.1109/ICCV48922.2021.00038","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00038","url":null,"abstract":"A standard practice of deploying deep neural networks is to apply the same architecture to all the input instances. However, a fixed architecture may not be suitable for different data with high diversity. To boost the model capacity, existing methods usually employ larger convolutional kernels or deeper network layers, which incurs prohibitive computational costs. In this paper, we address this issue by proposing Differentiable Dynamic Wirings (DDW), which learns the instance-aware connectivity that creates different wiring patterns for different instances. 1) Specifically, the network is initialized as a complete directed acyclic graph, where the nodes represent convolutional blocks and the edges represent the connection paths. 2) We generate edge weights by a learnable module, Router, and select the edges whose weights are larger than a threshold, to adjust the connectivity of the neural network structure. 3) Instead of using the same path of the network, DDW aggregates features dynamically in each node, which allows the network to have more representation power.To facilitate effective training, we further represent the network connectivity of each sample as an adjacency matrix. The matrix is updated to aggregate features in the forward pass, cached in the memory, and used for gradient computing in the backward pass. We validate the effectiveness of our approach with several mainstream architectures, including MobileNetV2, ResNet, ResNeXt, and RegNet. Extensive experiments are performed on ImageNet classification and COCO object detection, which demonstrates the effectiveness and generalization ability of our approach.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"3 1","pages":"317-326"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73796221","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 : 2021-10-01DOI: 10.1109/ICCV48922.2021.00642
Guile Wu, S. Gong
Contemporary domain generalization (DG) and multisource unsupervised domain adaptation (UDA) methods mostly collect data from multiple domains together for joint optimization. However, this centralized training paradigm poses a threat to data privacy and is not applicable when data are non-shared across domains. In this work, we propose a new approach called Collaborative Optimization and Aggregation (COPA), which aims at optimizing a generalized target model for decentralized DG and UDA, where data from different domains are non-shared and private. Our base model consists of a domain-invariant feature extractor and an ensemble of domain-specific classifiers. In an iterative learning process, we optimize a local model for each domain, and then centrally aggregate local feature extractors and assemble domain-specific classifiers to construct a generalized global model, without sharing data from different domains. To improve generalization of feature extractors, we employ hybrid batch-instance normalization and collaboration of frozen classifiers. For better decentralized UDA, we further introduce a prediction agreement mechanism to overcome local disparities towards central model aggregation. Extensive experiments on five DG and UDA benchmark datasets show that COPA is capable of achieving comparable performance against the state-of-the-art DG and UDA methods without the need for centralized data collection in model training.
{"title":"Collaborative Optimization and Aggregation for Decentralized Domain Generalization and Adaptation","authors":"Guile Wu, S. Gong","doi":"10.1109/ICCV48922.2021.00642","DOIUrl":"https://doi.org/10.1109/ICCV48922.2021.00642","url":null,"abstract":"Contemporary domain generalization (DG) and multisource unsupervised domain adaptation (UDA) methods mostly collect data from multiple domains together for joint optimization. However, this centralized training paradigm poses a threat to data privacy and is not applicable when data are non-shared across domains. In this work, we propose a new approach called Collaborative Optimization and Aggregation (COPA), which aims at optimizing a generalized target model for decentralized DG and UDA, where data from different domains are non-shared and private. Our base model consists of a domain-invariant feature extractor and an ensemble of domain-specific classifiers. In an iterative learning process, we optimize a local model for each domain, and then centrally aggregate local feature extractors and assemble domain-specific classifiers to construct a generalized global model, without sharing data from different domains. To improve generalization of feature extractors, we employ hybrid batch-instance normalization and collaboration of frozen classifiers. For better decentralized UDA, we further introduce a prediction agreement mechanism to overcome local disparities towards central model aggregation. Extensive experiments on five DG and UDA benchmark datasets show that COPA is capable of achieving comparable performance against the state-of-the-art DG and UDA methods without the need for centralized data collection in model training.","PeriodicalId":6820,"journal":{"name":"2021 IEEE/CVF International Conference on Computer Vision (ICCV)","volume":"30 1","pages":"6464-6473"},"PeriodicalIF":0.0,"publicationDate":"2021-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73940799","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：