Pub Date : 2020-10-21DOI: 10.1109/3DV50981.2020.00123
François Darmon, Mathieu Aubry, P. Monasse
We tackle the problem of finding accurate and robust keypoint correspondences between images. We propose a learning-based approach to guide local feature matches via a learned approximate image matching. Our approach can boost the results of SIFT to a level similar to state-of-the-art deep descriptors, such as Superpoint, ContextDesc, or D2-Net and can improve performance for these descriptors. We introduce and study different levels of supervision to learn coarse correspondences. In particular, we show that weak supervision from epipolar geometry leads to performances higher than the stronger but more biased point level supervision and is a clear improvement over weak image level supervision. We demonstrate the benefits of our approach in a variety of conditions by evaluating our guided keypoint correspondences for localization of internet images on the YFCC100M dataset and indoor images on the SUN3D dataset, for robust localization on the Aachen day-night benchmark and for 3D reconstruction in challenging conditions using the LTLL historical image data.
{"title":"Learning to Guide Local Feature Matches","authors":"François Darmon, Mathieu Aubry, P. Monasse","doi":"10.1109/3DV50981.2020.00123","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00123","url":null,"abstract":"We tackle the problem of finding accurate and robust keypoint correspondences between images. We propose a learning-based approach to guide local feature matches via a learned approximate image matching. Our approach can boost the results of SIFT to a level similar to state-of-the-art deep descriptors, such as Superpoint, ContextDesc, or D2-Net and can improve performance for these descriptors. We introduce and study different levels of supervision to learn coarse correspondences. In particular, we show that weak supervision from epipolar geometry leads to performances higher than the stronger but more biased point level supervision and is a clear improvement over weak image level supervision. We demonstrate the benefits of our approach in a variety of conditions by evaluating our guided keypoint correspondences for localization of internet images on the YFCC100M dataset and indoor images on the SUN3D dataset, for robust localization on the Aachen day-night benchmark and for 3D reconstruction in challenging conditions using the LTLL historical image data.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116057940","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 : 2020-10-21DOI: 10.1109/3DV50981.2020.00101
Felix Taubner, Florian Tschopp, Tonci Novkovic, R. Siegwart, Fadri Furrer
Current research on visual place recognition mostly focuses on aggregating local visual features of an image into a single vector representation. Therefore, high-level information such as the geometric arrangement of the features is typically lost. In this paper, we introduce a novel learning-based approach to place recognition, using RGB-D cameras and line clusters as visual and geometric features. We state the place recognition problem as a problem of recognizing clusters of lines instead of individual patches, thus maintaining structural information. In our work, line clusters are defined as lines that make up individual objects, hence our place recognition approach can be understood as object recognition. 3D line segments are detected in RGB-D images using state-of-the-art techniques. We present a neural network architecture based on the attention mechanism for frame-wise line clustering. A similar neural network is used for the description of these clusters with a compact embedding of 128 floating point numbers, trained with triplet loss on training data obtained from the InteriorNet dataset. We show experiments on a large number of indoor scenes and compare our method with the bag-of-words image-retrieval approach using SIFT and SuperPoint features and the global descriptor NetVLAD. Trained only on synthetic data, our approach generalizes well to real-world data captured with Kinect sensors, while also providing information about the geometric arrangement of instances.
{"title":"LCD – Line Clustering and Description for Place Recognition","authors":"Felix Taubner, Florian Tschopp, Tonci Novkovic, R. Siegwart, Fadri Furrer","doi":"10.1109/3DV50981.2020.00101","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00101","url":null,"abstract":"Current research on visual place recognition mostly focuses on aggregating local visual features of an image into a single vector representation. Therefore, high-level information such as the geometric arrangement of the features is typically lost. In this paper, we introduce a novel learning-based approach to place recognition, using RGB-D cameras and line clusters as visual and geometric features. We state the place recognition problem as a problem of recognizing clusters of lines instead of individual patches, thus maintaining structural information. In our work, line clusters are defined as lines that make up individual objects, hence our place recognition approach can be understood as object recognition. 3D line segments are detected in RGB-D images using state-of-the-art techniques. We present a neural network architecture based on the attention mechanism for frame-wise line clustering. A similar neural network is used for the description of these clusters with a compact embedding of 128 floating point numbers, trained with triplet loss on training data obtained from the InteriorNet dataset. We show experiments on a large number of indoor scenes and compare our method with the bag-of-words image-retrieval approach using SIFT and SuperPoint features and the global descriptor NetVLAD. Trained only on synthetic data, our approach generalizes well to real-world data captured with Kinect sensors, while also providing information about the geometric arrangement of instances.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132059247","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 : 2020-10-19DOI: 10.1109/3DV50981.2020.00113
Vinit Sarode, Animesh Dhagat, Rangaprasad Arun Srivatsan, N. Zevallos, S. Lucey, H. Choset
Point clouds have grown in importance in the way computers perceive the world. From LIDAR sensors in autonomous cars and drones to the time of flight and stereo vision systems in our phones, point clouds are everywhere. Despite their ubiquity, point clouds in the real world are often missing points because of sensor limitations or occlusions, or contain extraneous points from sensor noise or artifacts. These problems challenge algorithms that require computing correspondences between a pair of point clouds. Therefore, this paper presents a fully-convolutional neural network that identifies which points in one point cloud are most similar (inliers) to the points in another. We show improvements in learning-based and classical point cloud registration approaches when retrofitted with our network. We demonstrate these improvements on synthetic and real-world datasets. Finally, our network produces impressive results on test datasets that were unseen during training, thus exhibiting generalizability. Code and videos are available at https://github.com/vinits5/masknet
{"title":"MaskNet: A Fully-Convolutional Network to Estimate Inlier Points","authors":"Vinit Sarode, Animesh Dhagat, Rangaprasad Arun Srivatsan, N. Zevallos, S. Lucey, H. Choset","doi":"10.1109/3DV50981.2020.00113","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00113","url":null,"abstract":"Point clouds have grown in importance in the way computers perceive the world. From LIDAR sensors in autonomous cars and drones to the time of flight and stereo vision systems in our phones, point clouds are everywhere. Despite their ubiquity, point clouds in the real world are often missing points because of sensor limitations or occlusions, or contain extraneous points from sensor noise or artifacts. These problems challenge algorithms that require computing correspondences between a pair of point clouds. Therefore, this paper presents a fully-convolutional neural network that identifies which points in one point cloud are most similar (inliers) to the points in another. We show improvements in learning-based and classical point cloud registration approaches when retrofitted with our network. We demonstrate these improvements on synthetic and real-world datasets. Finally, our network produces impressive results on test datasets that were unseen during training, thus exhibiting generalizability. Code and videos are available at https://github.com/vinits5/masknet","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126908873","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 : 2020-10-19DOI: 10.1109/3DV50981.2020.00080
Nitin Agarwal, M. Gopi
We present a new meshing algorithm called guided and augmented meshing, GAMesh, which uses a mesh prior to generate a surface for the output points of a point network. By projecting the output points onto this prior and simplifying the resulting mesh, GAMesh ensures a surface with the same topology as the mesh prior but whose geometric fidelity is controlled by the point network. This makes GAMesh independent of both the density and distribution of the output points, a common artifact in traditional surface reconstruction algorithms. We show that such a separation of geometry from topology can have several advantages especially in single-view shape prediction, fair evaluation of point networks and reconstructing surfaces for networks which output sparse point clouds. We further show that by training point networks with GAMesh, we can directly optimize the vertex positions to generate adaptive meshes with arbitrary topologies. Code and data are available on the project webpage1.1https://www.ics.uci.edu/∼agarwal/GAMesh
{"title":"GAMesh: Guided and Augmented Meshing for Deep Point Networks","authors":"Nitin Agarwal, M. Gopi","doi":"10.1109/3DV50981.2020.00080","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00080","url":null,"abstract":"We present a new meshing algorithm called guided and augmented meshing, GAMesh, which uses a mesh prior to generate a surface for the output points of a point network. By projecting the output points onto this prior and simplifying the resulting mesh, GAMesh ensures a surface with the same topology as the mesh prior but whose geometric fidelity is controlled by the point network. This makes GAMesh independent of both the density and distribution of the output points, a common artifact in traditional surface reconstruction algorithms. We show that such a separation of geometry from topology can have several advantages especially in single-view shape prediction, fair evaluation of point networks and reconstructing surfaces for networks which output sparse point clouds. We further show that by training point networks with GAMesh, we can directly optimize the vertex positions to generate adaptive meshes with arbitrary topologies. Code and data are available on the project webpage1.1https://www.ics.uci.edu/∼agarwal/GAMesh","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126251775","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 : 2020-10-18DOI: 10.1109/3DV50981.2020.00034
Mahdi Saleh, Shervin Dehghani, Benjamin Busam, N. Navab, Federico Tombari
3D Point clouds are a rich source of information that enjoy growing popularity in the vision community. However, due to the sparsity of their representation, learning models based on large point clouds is still a challenge. In this work, we introduce Graphite, a GRAPH-Induced feaTure Extraction pipeline, a simple yet powerful feature transform and keypoint detector. Graphite enables intensive down-sampling of point clouds with keypoint detection accompanied by a descriptor. We construct a generic graph-based learning scheme to describe point cloud regions and extract salient points. To this end, we take advantage of 6D pose information and metric learning to learn robust descriptions and keypoints across different scans. We Reformulate the 3D keypoint pipeline with graph neural networks which allow efficient processing of the point set while boosting its descriptive power which ultimately results in more accurate 3D registrations. We demonstrate our lightweight descriptor on common 3D descriptor matching and point cloud registration benchmarks [76], [71] and achieve comparable results with the state of the art. Describing 100 patches of a point cloud and detecting their keypoints takes only 0.018 seconds with our proposed network.
{"title":"Graphite: Graph-Induced Feature Extraction for Point Cloud Registration","authors":"Mahdi Saleh, Shervin Dehghani, Benjamin Busam, N. Navab, Federico Tombari","doi":"10.1109/3DV50981.2020.00034","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00034","url":null,"abstract":"3D Point clouds are a rich source of information that enjoy growing popularity in the vision community. However, due to the sparsity of their representation, learning models based on large point clouds is still a challenge. In this work, we introduce Graphite, a GRAPH-Induced feaTure Extraction pipeline, a simple yet powerful feature transform and keypoint detector. Graphite enables intensive down-sampling of point clouds with keypoint detection accompanied by a descriptor. We construct a generic graph-based learning scheme to describe point cloud regions and extract salient points. To this end, we take advantage of 6D pose information and metric learning to learn robust descriptions and keypoints across different scans. We Reformulate the 3D keypoint pipeline with graph neural networks which allow efficient processing of the point set while boosting its descriptive power which ultimately results in more accurate 3D registrations. We demonstrate our lightweight descriptor on common 3D descriptor matching and point cloud registration benchmarks [76], [71] and achieve comparable results with the state of the art. Describing 100 patches of a point cloud and detecting their keypoints takes only 0.018 seconds with our proposed network.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"55 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114238433","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 : 2020-10-16DOI: 10.1109/3DV50981.2020.00063
Javier Hidalgo-Carri'o, Daniel Gehrig, D. Scaramuzza
Event cameras are novel sensors that output brightness changes in the form of a stream of asynchronous ”events” instead of intensity frames. Compared to conventional image sensors, they offer significant advantages: high temporal resolution, high dynamic range, no motion blur, and much lower bandwidth. Recently, learning-based approaches have been applied to event-based data, thus unlocking their potential and making significant progress in a variety of tasks, such as monocular depth prediction. Most existing approaches use standard feed-forward architectures to generate network predictions, which do not leverage the temporal consistency presents in the event stream. We propose a recurrent architecture to solve this task and show significant improvement over standard feed-forward methods. In particular, our method generates dense depth predictions using a monocular setup, which has not been shown previously. We pretrain our model using a new dataset containing events and depth maps recorded in the CARLA simulator. We test our method on the Multi Vehicle Stereo Event Camera Dataset (MVSEC). Quantitative experiments show up to 50% improvement in average depth error with respect to previous event-based methods. Code and dataset are available at: http://rpg.ifi.uzh.ch/e2depth
{"title":"Learning Monocular Dense Depth from Events","authors":"Javier Hidalgo-Carri'o, Daniel Gehrig, D. Scaramuzza","doi":"10.1109/3DV50981.2020.00063","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00063","url":null,"abstract":"Event cameras are novel sensors that output brightness changes in the form of a stream of asynchronous ”events” instead of intensity frames. Compared to conventional image sensors, they offer significant advantages: high temporal resolution, high dynamic range, no motion blur, and much lower bandwidth. Recently, learning-based approaches have been applied to event-based data, thus unlocking their potential and making significant progress in a variety of tasks, such as monocular depth prediction. Most existing approaches use standard feed-forward architectures to generate network predictions, which do not leverage the temporal consistency presents in the event stream. We propose a recurrent architecture to solve this task and show significant improvement over standard feed-forward methods. In particular, our method generates dense depth predictions using a monocular setup, which has not been shown previously. We pretrain our model using a new dataset containing events and depth maps recorded in the CARLA simulator. We test our method on the Multi Vehicle Stereo Event Camera Dataset (MVSEC). Quantitative experiments show up to 50% improvement in average depth error with respect to previous event-based methods. Code and dataset are available at: http://rpg.ifi.uzh.ch/e2depth","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125283922","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 : 2020-10-14DOI: 10.1109/3DV50981.2020.00069
Zhantao Deng, Jan Bednarík, M. Salzmann, P. Fua
Recently, parametric mappings have emerged as highly effective surface representations, yielding low reconstruction error. In particular, the latest works represent the target shape as an atlas of multiple mappings, which can closely encode object parts. Atlas representations, however, suffer from one major drawback: The individual mappings are not guaranteed to be consistent, which results in holes in the reconstructed shape or in jagged surface areas.We introduce an approach that explicitly encourages global consistency of the local mappings. To this end, we introduce two novel loss terms. The first term exploits the surface normals and requires that they remain locally consistent when estimated within and across the individual mappings. The second term further encourages better spatial configuration of the mappings by minimizing novel stitching error. We show on standard benchmarks that the use of normal consistency requirement outperforms the baselines quantitatively while enforcing better stitching leads to much better visual quality of the reconstructed objects as compared to the state-of-the-art.
{"title":"Better Patch Stitching for Parametric Surface Reconstruction","authors":"Zhantao Deng, Jan Bednarík, M. Salzmann, P. Fua","doi":"10.1109/3DV50981.2020.00069","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00069","url":null,"abstract":"Recently, parametric mappings have emerged as highly effective surface representations, yielding low reconstruction error. In particular, the latest works represent the target shape as an atlas of multiple mappings, which can closely encode object parts. Atlas representations, however, suffer from one major drawback: The individual mappings are not guaranteed to be consistent, which results in holes in the reconstructed shape or in jagged surface areas.We introduce an approach that explicitly encourages global consistency of the local mappings. To this end, we introduce two novel loss terms. The first term exploits the surface normals and requires that they remain locally consistent when estimated within and across the individual mappings. The second term further encourages better spatial configuration of the mappings by minimizing novel stitching error. We show on standard benchmarks that the use of normal consistency requirement outperforms the baselines quantitatively while enforcing better stitching leads to much better visual quality of the reconstructed objects as compared to the state-of-the-art.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115552986","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 : 2020-10-14DOI: 10.1109/3DV50981.2020.00047
Changjiang Cai, Philippos Mordohai
Deep networks for stereo matching typically leverage 2D or 3D convolutional encoder-decoder architectures to aggregate cost and regularize the cost volume for accurate disparity estimation. Due to content-insensitive convolutions and down-sampling and up-sampling operations, these cost aggregation mechanisms do not take full advantage of the information available in the images. Disparity maps suffer from over-smoothing near occlusion boundaries, and erroneous predictions in thin structures. In this paper, we show how deep adaptive filtering and differentiable semi-global aggregation can be integrated in existing 2D and 3D convolutional networks for end-to-end stereo matching, leading to improved accuracy. The improvements are due to utilizing RGB information from the images as a signal to dynamically guide the matching process, in addition to being the signal we attempt to match across the images. We show extensive experimental results on the KITTI 2015 and Virtual KITTI 2 datasets comparing four stereo networks (DispNetC, GCNet, PSMNet and GANet) after integrating four adaptive filters (segmentation-aware bilateral filtering, dynamic filtering networks, pixel adaptive convolution and semi-global aggregation) into their architectures. Our code is available at https://github.com/ccj5351/DAFStereoNets.
{"title":"Do End-to-end Stereo Algorithms Under-utilize Information?","authors":"Changjiang Cai, Philippos Mordohai","doi":"10.1109/3DV50981.2020.00047","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00047","url":null,"abstract":"Deep networks for stereo matching typically leverage 2D or 3D convolutional encoder-decoder architectures to aggregate cost and regularize the cost volume for accurate disparity estimation. Due to content-insensitive convolutions and down-sampling and up-sampling operations, these cost aggregation mechanisms do not take full advantage of the information available in the images. Disparity maps suffer from over-smoothing near occlusion boundaries, and erroneous predictions in thin structures. In this paper, we show how deep adaptive filtering and differentiable semi-global aggregation can be integrated in existing 2D and 3D convolutional networks for end-to-end stereo matching, leading to improved accuracy. The improvements are due to utilizing RGB information from the images as a signal to dynamically guide the matching process, in addition to being the signal we attempt to match across the images. We show extensive experimental results on the KITTI 2015 and Virtual KITTI 2 datasets comparing four stereo networks (DispNetC, GCNet, PSMNet and GANet) after integrating four adaptive filters (segmentation-aware bilateral filtering, dynamic filtering networks, pixel adaptive convolution and semi-global aggregation) into their architectures. Our code is available at https://github.com/ccj5351/DAFStereoNets.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"247 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114605952","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 : 2020-10-14DOI: 10.1109/3DV50981.2020.00118
Alican Mertan, Y. Sahin, D. Duff, Gözde B. Ünal
We propose a new approach for the problem of relative depth estimation from a single image. Instead of directly regressing over depth scores, we formulate the problem as estimation of a probability distribution over depth and aim to learn the parameters of the distributions which maximize the likelihood of the given data. To train our model, we propose a new ranking loss, Distributional Loss, which tries to increase the probability of farther pixel’s depth being greater than the closer pixel’s depth. Our proposed approach allows our model to output confidence in its estimation in the form of standard deviation of the distribution. We achieve state of the art results against a number of baselines while providing confidence in our estimations. Our analysis show that estimated confidence is actually a good indicator of accuracy. We investigate the usage of confidence information in a downstream task of metric depth estimation, to increase its performance.
{"title":"A New Distributional Ranking Loss With Uncertainty: Illustrated in Relative Depth Estimation","authors":"Alican Mertan, Y. Sahin, D. Duff, Gözde B. Ünal","doi":"10.1109/3DV50981.2020.00118","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00118","url":null,"abstract":"We propose a new approach for the problem of relative depth estimation from a single image. Instead of directly regressing over depth scores, we formulate the problem as estimation of a probability distribution over depth and aim to learn the parameters of the distributions which maximize the likelihood of the given data. To train our model, we propose a new ranking loss, Distributional Loss, which tries to increase the probability of farther pixel’s depth being greater than the closer pixel’s depth. Our proposed approach allows our model to output confidence in its estimation in the form of standard deviation of the distribution. We achieve state of the art results against a number of baselines while providing confidence in our estimations. Our analysis show that estimated confidence is actually a good indicator of accuracy. We investigate the usage of confidence information in a downstream task of metric depth estimation, to increase its performance.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121768730","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 : 2020-10-14DOI: 10.1109/3DV50981.2020.00046
Changjiang Cai, Matteo Poggi, S. Mattoccia, Philippos Mordohai
End-to-end deep networks represent the state of the art for stereo matching. While excelling on images framing environments similar to the training set, major drops in accuracy occur in unseen domains (e.g., when moving from synthetic to real scenes). In this paper we introduce a novel family of architectures, namely Matching-Space Networks (MS-Nets), with improved generalization properties. By replacing learning-based feature extraction from image RGB values with matching functions and confidence measures from conventional wisdom, we move the learning process from the color space to the Matching Space, avoiding over-specialization to domain specific features. Extensive experimental results on four real datasets highlight that our proposal leads to superior generalization to unseen environments over conventional deep architectures, keeping accuracy on the source domain almost unaltered. Our code is available at https://qithub.com/ccj5351/MS-Nets.
{"title":"Matching-space Stereo Networks for Cross-domain Generalization","authors":"Changjiang Cai, Matteo Poggi, S. Mattoccia, Philippos Mordohai","doi":"10.1109/3DV50981.2020.00046","DOIUrl":"https://doi.org/10.1109/3DV50981.2020.00046","url":null,"abstract":"End-to-end deep networks represent the state of the art for stereo matching. While excelling on images framing environments similar to the training set, major drops in accuracy occur in unseen domains (e.g., when moving from synthetic to real scenes). In this paper we introduce a novel family of architectures, namely Matching-Space Networks (MS-Nets), with improved generalization properties. By replacing learning-based feature extraction from image RGB values with matching functions and confidence measures from conventional wisdom, we move the learning process from the color space to the Matching Space, avoiding over-specialization to domain specific features. Extensive experimental results on four real datasets highlight that our proposal leads to superior generalization to unseen environments over conventional deep architectures, keeping accuracy on the source domain almost unaltered. Our code is available at https://qithub.com/ccj5351/MS-Nets.","PeriodicalId":293399,"journal":{"name":"2020 International Conference on 3D Vision (3DV)","volume":"88 11","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133356002","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：