Pub Date : 2020-06-01DOI: 10.1109/cvpr42600.2020.00213
Shaifali Parashar, M. Salzmann, P. Fua
We propose a new formulation to non-rigid structure-from-motion that only requires the deforming surface to preserve its differential structure. This is a much weaker assumption than the traditional ones of isometry or conformality. We show that it is nevertheless sufficient to establish local correspondences between the surface in two different images and therefore to perform point-wise reconstruction using only first-order derivatives. To this end, we formulate differential constraints and solve them algebraically using the theory of resultants. We will demonstrate that our approach is more widely applicable, more stable in noisy and sparse imaging conditions and much faster than earlier ones, while delivering similar accuracy. The code is available at https://github.com/cvlab-epfl/diff-nrsfm/.
{"title":"Local Non-Rigid Structure-From-Motion From Diffeomorphic Mappings","authors":"Shaifali Parashar, M. Salzmann, P. Fua","doi":"10.1109/cvpr42600.2020.00213","DOIUrl":"https://doi.org/10.1109/cvpr42600.2020.00213","url":null,"abstract":"We propose a new formulation to non-rigid structure-from-motion that only requires the deforming surface to preserve its differential structure. This is a much weaker assumption than the traditional ones of isometry or conformality. We show that it is nevertheless sufficient to establish local correspondences between the surface in two different images and therefore to perform point-wise reconstruction using only first-order derivatives. To this end, we formulate differential constraints and solve them algebraically using the theory of resultants. We will demonstrate that our approach is more widely applicable, more stable in noisy and sparse imaging conditions and much faster than earlier ones, while delivering similar accuracy. The code is available at https://github.com/cvlab-epfl/diff-nrsfm/.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"48 1","pages":"2056-2064"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90579185","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 : 2020-06-01DOI: 10.1109/CVPR42600.2020.00845
Gyumin Shim, Jinsun Park, I. Kweon
In this paper, we propose a novel and efficient reference feature extraction module referred to as the Similarity Search and Extraction Network (SSEN) for reference-based super-resolution (RefSR) tasks. The proposed module extracts aligned relevant features from a reference image to increase the performance over single image super-resolution (SISR) methods. In contrast to conventional algorithms which utilize brute-force searches or optical flow estimations, the proposed algorithm is end-to-end trainable without any additional supervision or heavy computation, predicting the best match with a single network forward operation. Moreover, the proposed module is aware of not only the best matching position but also the relevancy of the best match. This makes our algorithm substantially robust when irrelevant reference images are given, overcoming the major cause of the performance degradation when using existing RefSR methods. Furthermore, our module can be utilized for self-similarity SR if no reference image is available. Experimental results demonstrate the superior performance of the proposed algorithm compared to previous works both quantitatively and qualitatively.
{"title":"Robust Reference-Based Super-Resolution With Similarity-Aware Deformable Convolution","authors":"Gyumin Shim, Jinsun Park, I. Kweon","doi":"10.1109/CVPR42600.2020.00845","DOIUrl":"https://doi.org/10.1109/CVPR42600.2020.00845","url":null,"abstract":"In this paper, we propose a novel and efficient reference feature extraction module referred to as the Similarity Search and Extraction Network (SSEN) for reference-based super-resolution (RefSR) tasks. The proposed module extracts aligned relevant features from a reference image to increase the performance over single image super-resolution (SISR) methods. In contrast to conventional algorithms which utilize brute-force searches or optical flow estimations, the proposed algorithm is end-to-end trainable without any additional supervision or heavy computation, predicting the best match with a single network forward operation. Moreover, the proposed module is aware of not only the best matching position but also the relevancy of the best match. This makes our algorithm substantially robust when irrelevant reference images are given, overcoming the major cause of the performance degradation when using existing RefSR methods. Furthermore, our module can be utilized for self-similarity SR if no reference image is available. Experimental results demonstrate the superior performance of the proposed algorithm compared to previous works both quantitatively and qualitatively.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"85 3 1","pages":"8422-8431"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90633524","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}
Existing weakly-supervised semantic segmentation methods using image-level annotations typically rely on initial responses to locate object regions. However, such response maps generated by the classification network usually focus on discriminative object parts, due to the fact that the network does not need the entire object for optimizing the objective function. To enforce the network to pay attention to other parts of an object, we propose a simple yet effective approach that introduces a self-supervised task by exploiting the sub-category information. Specifically, we perform clustering on image features to generate pseudo sub-categories labels within each annotated parent class, and construct a sub-category objective to assign the network to a more challenging task. By iteratively clustering image features, the training process does not limit itself to the most discriminative object parts, hence improving the quality of the response maps. We conduct extensive analysis to validate the proposed method and show that our approach performs favorably against the state-of-the-art approaches.
{"title":"Weakly-Supervised Semantic Segmentation via Sub-Category Exploration","authors":"Yu-Ting Chang, Qiaosong Wang, Wei-Chih Hung, Robinson Piramuthu, Yi-Hsuan Tsai, Ming-Hsuan Yang","doi":"10.1109/cvpr42600.2020.00901","DOIUrl":"https://doi.org/10.1109/cvpr42600.2020.00901","url":null,"abstract":"Existing weakly-supervised semantic segmentation methods using image-level annotations typically rely on initial responses to locate object regions. However, such response maps generated by the classification network usually focus on discriminative object parts, due to the fact that the network does not need the entire object for optimizing the objective function. To enforce the network to pay attention to other parts of an object, we propose a simple yet effective approach that introduces a self-supervised task by exploiting the sub-category information. Specifically, we perform clustering on image features to generate pseudo sub-categories labels within each annotated parent class, and construct a sub-category objective to assign the network to a more challenging task. By iteratively clustering image features, the training process does not limit itself to the most discriminative object parts, hence improving the quality of the response maps. We conduct extensive analysis to validate the proposed method and show that our approach performs favorably against the state-of-the-art approaches.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"14 1","pages":"8988-8997"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90997304","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 : 2020-06-01DOI: 10.1109/CVPR42600.2020.01026
Muli Yang, Cheng Deng, Junchi Yan, Xianglong Liu, D. Tao
Composing and recognizing new concepts from known sub-concepts has been a fundamental and challenging vision task, mainly due to 1) the diversity of sub-concepts and 2) the intricate contextuality between sub-concepts and their corresponding visual features. However, most of the current methods simply treat the contextuality as rigid semantic relationships and fail to capture fine-grained contextual correlations. We propose to learn unseen concepts in a hierarchical decomposition-and-composition manner. Considering the diversity of sub-concepts, our method decomposes each seen image into visual elements according to its labels, and learns corresponding sub-concepts in their individual subspaces. To model intricate contextuality between sub-concepts and their visual features, compositions are generated from these subspaces in three hierarchical forms, and the composed concepts are learned in a unified composition space. To further refine the captured contextual relationships, adaptively semi-positive concepts are defined and then learned with pseudo supervision exploited from the generated compositions. We validate the proposed approach on two challenging benchmarks, and demonstrate its superiority over state-of-the-art approaches.
{"title":"Learning Unseen Concepts via Hierarchical Decomposition and Composition","authors":"Muli Yang, Cheng Deng, Junchi Yan, Xianglong Liu, D. Tao","doi":"10.1109/CVPR42600.2020.01026","DOIUrl":"https://doi.org/10.1109/CVPR42600.2020.01026","url":null,"abstract":"Composing and recognizing new concepts from known sub-concepts has been a fundamental and challenging vision task, mainly due to 1) the diversity of sub-concepts and 2) the intricate contextuality between sub-concepts and their corresponding visual features. However, most of the current methods simply treat the contextuality as rigid semantic relationships and fail to capture fine-grained contextual correlations. We propose to learn unseen concepts in a hierarchical decomposition-and-composition manner. Considering the diversity of sub-concepts, our method decomposes each seen image into visual elements according to its labels, and learns corresponding sub-concepts in their individual subspaces. To model intricate contextuality between sub-concepts and their visual features, compositions are generated from these subspaces in three hierarchical forms, and the composed concepts are learned in a unified composition space. To further refine the captured contextual relationships, adaptively semi-positive concepts are defined and then learned with pseudo supervision exploited from the generated compositions. We validate the proposed approach on two challenging benchmarks, and demonstrate its superiority over state-of-the-art approaches.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"144 1","pages":"10245-10253"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89766060","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}
Recently, extensive researches have been proposed to address the UDA problem, which aims to learn transferrable models for the unlabeled target domain. Among them, the optimal transport is a promising metric to align the representations of the source and target domains. However, most existing works based on optimal transport ignore the intra-domain structure, only achieving coarse pair-wise matching. The target samples distributed near the edge of the clusters, or far from their corresponding class centers are easily to be misclassified by the decision boundary learned from the source domain. In this paper, we present Reliable Weighted Optimal Transport (RWOT) for unsupervised domain adaptation, including novel Shrinking Subspace Reliability (SSR) and weighted optimal transport strategy. Specifically, SSR exploits spatial prototypical information and intra-domain structure to dynamically measure the sample-level domain discrepancy across domains. Besides, the weighted optimal transport strategy based on SSR is exploited to achieve the precise-pair-wise optimal transport procedure, which reduces negative transfer brought by the samples near decision boundaries in the target domain. RWOT also equips with the discriminative centroid clustering exploitation strategy to learn transfer features. A thorough evaluation shows that RWOT outperforms existing state-of-the-art method on standard domain adaptation benchmarks.
{"title":"Reliable Weighted Optimal Transport for Unsupervised Domain Adaptation","authors":"Renjun Xu, Pelen Liu, Liyan Wang, Chao Chen, Jindong Wang, Kaiming He, X. Zhang, Shaoqing Ren, Mingsheng Long, Zhangjie Cao, Jianmin Wang","doi":"10.1109/cvpr42600.2020.00445","DOIUrl":"https://doi.org/10.1109/cvpr42600.2020.00445","url":null,"abstract":"Recently, extensive researches have been proposed to address the UDA problem, which aims to learn transferrable models for the unlabeled target domain. Among them, the optimal transport is a promising metric to align the representations of the source and target domains. However, most existing works based on optimal transport ignore the intra-domain structure, only achieving coarse pair-wise matching. The target samples distributed near the edge of the clusters, or far from their corresponding class centers are easily to be misclassified by the decision boundary learned from the source domain. In this paper, we present Reliable Weighted Optimal Transport (RWOT) for unsupervised domain adaptation, including novel Shrinking Subspace Reliability (SSR) and weighted optimal transport strategy. Specifically, SSR exploits spatial prototypical information and intra-domain structure to dynamically measure the sample-level domain discrepancy across domains. Besides, the weighted optimal transport strategy based on SSR is exploited to achieve the precise-pair-wise optimal transport procedure, which reduces negative transfer brought by the samples near decision boundaries in the target domain. RWOT also equips with the discriminative centroid clustering exploitation strategy to learn transfer features. A thorough evaluation shows that RWOT outperforms existing state-of-the-art method on standard domain adaptation benchmarks.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"31 1","pages":"4393-4402"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87621400","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}
Existing weakly supervised fine-grained image recognition (WFGIR) methods usually pick out the discriminative regions from the high-level feature maps directly. We discover that due to the operation of stacking local receptive filed, Convolutional Neural Network causes the discriminative region diffusion in high-level feature maps, which leads to inaccurate discriminative region localization. In this paper, we propose an end-to-end Discriminative Feature-oriented Gaussian Mixture Model (DF-GMM), to address the problem of discriminative region diffusion and find better fine-grained details. Specifically, DF-GMM consists of 1) a low-rank representation mechanism (LRM), which learns a set of low-rank discriminative bases by Gaussian Mixture Model (GMM) in high-level semantic feature maps to improve discriminative ability of feature representation, 2) a low-rank representation reorganization mechanism (LR$ ^2 $M) which resumes the space information corresponding to low-rank discriminative bases to reconstruct the low-rank feature maps. It alleviates the discriminative region diffusion problem and locate discriminative regions more precisely. Extensive experiments verify that DF-GMM yields the best performance under the same settings with the most competitive approaches, in CUB-Bird, Stanford-Cars datasets, and FGVC Aircraft.
{"title":"Weakly Supervised Fine-Grained Image Classification via Guassian Mixture Model Oriented Discriminative Learning","authors":"Zhihui Wang, Shijie Wang, Shuhui Yang, Haojie Li, Jianjun Li, Zezhou Li","doi":"10.1109/cvpr42600.2020.00977","DOIUrl":"https://doi.org/10.1109/cvpr42600.2020.00977","url":null,"abstract":"Existing weakly supervised fine-grained image recognition (WFGIR) methods usually pick out the discriminative regions from the high-level feature maps directly. We discover that due to the operation of stacking local receptive filed, Convolutional Neural Network causes the discriminative region diffusion in high-level feature maps, which leads to inaccurate discriminative region localization. In this paper, we propose an end-to-end Discriminative Feature-oriented Gaussian Mixture Model (DF-GMM), to address the problem of discriminative region diffusion and find better fine-grained details. Specifically, DF-GMM consists of 1) a low-rank representation mechanism (LRM), which learns a set of low-rank discriminative bases by Gaussian Mixture Model (GMM) in high-level semantic feature maps to improve discriminative ability of feature representation, 2) a low-rank representation reorganization mechanism (LR$ ^2 $M) which resumes the space information corresponding to low-rank discriminative bases to reconstruct the low-rank feature maps. It alleviates the discriminative region diffusion problem and locate discriminative regions more precisely. Extensive experiments verify that DF-GMM yields the best performance under the same settings with the most competitive approaches, in CUB-Bird, Stanford-Cars datasets, and FGVC Aircraft.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"27 1","pages":"9746-9755"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87903683","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 : 2020-06-01DOI: 10.1109/CVPR42600.2020.00423
Wanyu Lin, Zhaolin Gao, Baochun Li
Graph-based semi-supervised learning has been shown to be one of the most effective classification approaches, as it can exploit connectivity patterns between labeled and unlabeled samples to improve learning performance. However, we show that existing techniques perform poorly when labeled data are severely limited. To address the problem of semi-supervised learning in the presence of severely limited labeled samples, we propose a new framework, called {em Shoestring}, that incorporates metric learning into the paradigm of graph-based semi-supervised learning. In particular, our base model consists of a graph embedding network, followed by a metric learning network that learns a semantic metric space to represent the semantic similarity between the sparsely labeled and large numbers of unlabeled samples. Then the classification can be performed by clustering the unlabeled samples according to the learned semantic space. We empirically demonstrate Shoestring's superiority over many baselines, including graph convolutional networks, label propagation and their recent label-efficient variations (IGCN and GLP). We show that our framework achieves state-of-the-art performance for node classification in the low-data regime. In addition, we demonstrate the effectiveness of our framework on image classification tasks in the few-shot learning regime, with significant gains on miniImageNet ($2.57%sim3.59%$) and tieredImageNet ($1.05%sim2.70%$).
{"title":"Shoestring: Graph-Based Semi-Supervised Classification With Severely Limited Labeled Data","authors":"Wanyu Lin, Zhaolin Gao, Baochun Li","doi":"10.1109/CVPR42600.2020.00423","DOIUrl":"https://doi.org/10.1109/CVPR42600.2020.00423","url":null,"abstract":"Graph-based semi-supervised learning has been shown to be one of the most effective classification approaches, as it can exploit connectivity patterns between labeled and unlabeled samples to improve learning performance. However, we show that existing techniques perform poorly when labeled data are severely limited. To address the problem of semi-supervised learning in the presence of severely limited labeled samples, we propose a new framework, called {em Shoestring}, that incorporates metric learning into the paradigm of graph-based semi-supervised learning. In particular, our base model consists of a graph embedding network, followed by a metric learning network that learns a semantic metric space to represent the semantic similarity between the sparsely labeled and large numbers of unlabeled samples. Then the classification can be performed by clustering the unlabeled samples according to the learned semantic space. We empirically demonstrate Shoestring's superiority over many baselines, including graph convolutional networks, label propagation and their recent label-efficient variations (IGCN and GLP). We show that our framework achieves state-of-the-art performance for node classification in the low-data regime. In addition, we demonstrate the effectiveness of our framework on image classification tasks in the few-shot learning regime, with significant gains on miniImageNet ($2.57%sim3.59%$) and tieredImageNet ($1.05%sim2.70%$).","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"70 1","pages":"4173-4181"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86273122","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 : 2020-06-01DOI: 10.1109/CVPR42600.2020.00205
Fumihiko Sakaue, J. Sato
In this paper, we propose a method of visualizing 3D motion with zero latency. This method achieves motion visualization by projecting special high-frequency light patterns on moving objects without using any feedback mechanisms. For this objective, we focus on the time integration of light rays in the sensing system of observers. It is known that the visual system of human observers integrates light rays in a certain period. Similarly, the image sensor in a camera integrates light rays during the exposure time. Thus, our method embeds multiple images into a time-varying light field, such that the observer of the time-varying light field observes completely different images according to the dynamic motion of the scene. Based on this concept, we propose a method of generating special high-frequency patterns of projector lights. After projection onto target objects with projectors, the image observed on the target changes automatically depending on the motion of the objects and without any scene sensing and data analysis. In other words, we achieve motion visualization without the time delay incurred during sensing and computing.
{"title":"Active 3D Motion Visualization Based on Spatiotemporal Light-Ray Integration","authors":"Fumihiko Sakaue, J. Sato","doi":"10.1109/CVPR42600.2020.00205","DOIUrl":"https://doi.org/10.1109/CVPR42600.2020.00205","url":null,"abstract":"In this paper, we propose a method of visualizing 3D motion with zero latency. This method achieves motion visualization by projecting special high-frequency light patterns on moving objects without using any feedback mechanisms. For this objective, we focus on the time integration of light rays in the sensing system of observers. It is known that the visual system of human observers integrates light rays in a certain period. Similarly, the image sensor in a camera integrates light rays during the exposure time. Thus, our method embeds multiple images into a time-varying light field, such that the observer of the time-varying light field observes completely different images according to the dynamic motion of the scene. Based on this concept, we propose a method of generating special high-frequency patterns of projector lights. After projection onto target objects with projectors, the image observed on the target changes automatically depending on the motion of the objects and without any scene sensing and data analysis. In other words, we achieve motion visualization without the time delay incurred during sensing and computing.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"66 1","pages":"1977-1985"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86273408","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 : 2020-06-01DOI: 10.1109/cvpr42600.2020.00465
Shuai Bai, Zhiqun He, Y. Qiao, Hanzhe Hu, Wei Wu, Junjie Yan
The counting problem aims to estimate the number of objects in images. Due to large scale variation and labeling deviations, it remains a challenging task. The static density map supervised learning framework is widely used in existing methods, which uses the Gaussian kernel to generate a density map as the learning target and utilizes the Euclidean distance to optimize the model. However, the framework is intolerable to the labeling deviations and can not reflect the scale variation. In this paper, we propose an adaptive dilated convolution and a novel supervised learning framework named self-correction (SC) supervision. In the supervision level, the SC supervision utilizes the outputs of the model to iteratively correct the annotations and employs the SC loss to simultaneously optimize the model from both the whole and the individuals. In the feature level, the proposed adaptive dilated convolution predicts a continuous value as the specific dilation rate for each location, which adapts the scale variation better than a discrete and static dilation rate. Extensive experiments illustrate that our approach has achieved a consistent improvement on four challenging benchmarks. Especially, our approach achieves better performance than the state-of-the-art methods on all benchmark datasets.
{"title":"Adaptive Dilated Network With Self-Correction Supervision for Counting","authors":"Shuai Bai, Zhiqun He, Y. Qiao, Hanzhe Hu, Wei Wu, Junjie Yan","doi":"10.1109/cvpr42600.2020.00465","DOIUrl":"https://doi.org/10.1109/cvpr42600.2020.00465","url":null,"abstract":"The counting problem aims to estimate the number of objects in images. Due to large scale variation and labeling deviations, it remains a challenging task. The static density map supervised learning framework is widely used in existing methods, which uses the Gaussian kernel to generate a density map as the learning target and utilizes the Euclidean distance to optimize the model. However, the framework is intolerable to the labeling deviations and can not reflect the scale variation. In this paper, we propose an adaptive dilated convolution and a novel supervised learning framework named self-correction (SC) supervision. In the supervision level, the SC supervision utilizes the outputs of the model to iteratively correct the annotations and employs the SC loss to simultaneously optimize the model from both the whole and the individuals. In the feature level, the proposed adaptive dilated convolution predicts a continuous value as the specific dilation rate for each location, which adapts the scale variation better than a discrete and static dilation rate. Extensive experiments illustrate that our approach has achieved a consistent improvement on four challenging benchmarks. Especially, our approach achieves better performance than the state-of-the-art methods on all benchmark datasets.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"32 1","pages":"4593-4602"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86330858","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 : 2020-06-01DOI: 10.1109/cvpr42600.2020.00449
Yikai Li, Jiayuan Mao, Xiuming Zhang, W. Freeman, J. Tenenbaum, Jiajun Wu
We study the inverse graphics problem of inferring a holistic representation for natural images. Given an input image, our goal is to induce a neuro-symbolic, program-like representation that jointly models camera poses, object locations, and global scene structures. Such high-level, holistic scene representations further facilitate low-level image manipulation tasks such as inpainting. We formulate this problem as jointly finding the camera pose and scene structure that best describe the input image. The benefits of such joint inference are two-fold: scene regularity serves as a new cue for perspective correction, and in turn, correct perspective correction leads to a simplified scene structure, similar to how the correct shape leads to the most regular texture in shape from texture. Our proposed framework, Perspective Plane Program Induction (P3I), combines search-based and gradient-based algorithms to efficiently solve the problem. P3I outperforms a set of baselines on a collection of Internet images, across tasks including camera pose estimation, global structure inference, and down-stream image manipulation tasks.
{"title":"Perspective Plane Program Induction From a Single Image","authors":"Yikai Li, Jiayuan Mao, Xiuming Zhang, W. Freeman, J. Tenenbaum, Jiajun Wu","doi":"10.1109/cvpr42600.2020.00449","DOIUrl":"https://doi.org/10.1109/cvpr42600.2020.00449","url":null,"abstract":"We study the inverse graphics problem of inferring a holistic representation for natural images. Given an input image, our goal is to induce a neuro-symbolic, program-like representation that jointly models camera poses, object locations, and global scene structures. Such high-level, holistic scene representations further facilitate low-level image manipulation tasks such as inpainting. We formulate this problem as jointly finding the camera pose and scene structure that best describe the input image. The benefits of such joint inference are two-fold: scene regularity serves as a new cue for perspective correction, and in turn, correct perspective correction leads to a simplified scene structure, similar to how the correct shape leads to the most regular texture in shape from texture. Our proposed framework, Perspective Plane Program Induction (P3I), combines search-based and gradient-based algorithms to efficiently solve the problem. P3I outperforms a set of baselines on a collection of Internet images, across tasks including camera pose estimation, global structure inference, and down-stream image manipulation tasks.","PeriodicalId":6715,"journal":{"name":"2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","volume":"11 1","pages":"4433-4442"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86346330","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：