Filters








4,185 Hits in 5.0 sec

Joint Task-Recursive Learning for Semantic Segmentation and Depth Estimation [chapter]

Zhenyu Zhang, Zhen Cui, Chunyan Xu, Zequn Jie, Xiang Li, Jian Yang
2018 Lecture Notes in Computer Science  
In this paper, we propose a novel joint Task-Recursive Learning (TRL) framework for the closing-loop semantic segmentation and monocular depth estimation tasks.  ...  Extensive experiments on NYU-Depth v2 and SUN RGB-D datasets demonstrate that our method achieves state-of-the-art results for monocular depth estimation and semantic segmentation.  ...  In summary, the contributions of this paper are three folds: -Propose a novel joint Task-Recursive Learning (TRL) framework for semantic segmentation and depth estimation.  ... 
doi:10.1007/978-3-030-01249-6_15 fatcat:aikjy2xbsjabjfczjo2vwuspdi

Pattern-Structure Diffusion for Multi-Task Learning

Ling Zhou, Zhen Cui, Chunyan Xu, Zhenyu Zhang, Chaoqun Wang, Tong Zhang, Jian Yang
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
and task-across pattern structures in the task-level space for joint depth estimation, segmentation and surface normal prediction.  ...  Finally, the intra-task and inter-task pattern structures are jointly diffused among the task-level patterns, and encapsulated into an end-to-end PSD network to boost the performance of multi-task learning  ...  Acknowledgement This work was supported by the National Natural Science Foundation of China (Grants Nos.61772276, 61906094, 61972204, U1713208) and the Natural Science Foundation of Jiangsu Province (Grants  ... 
doi:10.1109/cvpr42600.2020.00457 dblp:conf/cvpr/ZhouCXZWZ020 fatcat:cgolbmcyezhyvgowjxiwa4j2ou

Latent 3D Volume for Joint Depth Estimation and Semantic Segmentation from a Single Image

Seiya Ito, Naoshi Kaneko, Kazuhiko Sumi
2020 Sensors  
This paper proposes a novel 3D representation, namely, a latent 3D volume, for joint depth estimation and semantic segmentation.  ...  We will show that the latent 3D volume is beneficial to the tasks of depth estimation and semantic segmentation because these tasks require an understanding of the 3D structure of the scene.  ...  Acknowledgments: Computational resources were provided by Aoyama Gakuin University Center for Advanced Information technology Research (CAIR).  ... 
doi:10.3390/s20205765 pmid:33053692 fatcat:mraupjui5ras5fswgoh3cunnri

Learning Semantic Segmentation from Synthetic Data: A Geometrically Guided Input-Output Adaptation Approach [article]

Yuhua Chen, Wen Li, Xiaoran Chen, Luc Van Gool
2019 arXiv   pre-print
network which simultaneously performs depth estimation and semantic segmentation on the synthetic data.  ...  This demonstrates the usefulness of geometric information from synthetic data for cross-domain semantic segmentation.  ...  : individually aligning both semantic segmentation and depth estimation; joint: aligning the joint output space of depth estimation and semantic segmentation, which is also our final output level adaptation  ... 
arXiv:1812.05040v2 fatcat:vp3wnkuoaff5hpgoy7htc32zfq

Competitive Simplicity for Multi-Task Learning for Real-Time Foggy Scene Understanding via Domain Adaptation [article]

Naif Alshammari, Samet Akcay, Toby P. Breckon
2020 arXiv   pre-print
In this paper, we propose a complex but competitive multi-task learning approach capable of performing in real-time semantic scene understanding and monocular depth estimation under foggy weather conditions  ...  For optimal performance in semantic segmentation, our model generates depth to be used as complementary source information with RGB in the segmentation network.  ...  As an overall loss for the depth estimation task, a joint depth loss is defined as follows: L jointdepth = L depth + L adv . (6) E.  ... 
arXiv:2012.05304v1 fatcat:irpbfthy4rgbfh7tmpaw3vjlfi

Towards Scene Understanding: Unsupervised Monocular Depth Estimation With Semantic-Aware Representation

Po-Yi Chen, Alexander H. Liu, Yen-Cheng Liu, Yu-Chiang Frank Wang
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Monocular depth estimation is a challenging task in scene understanding, with the goal to acquire the geometric properties of 3D space from 2D images.  ...  Moreover, our proposed model is able to perform region-aware depth estimation by enforcing semantics consistency between stereo pairs.  ...  This work is supported in part by the Ministry of Science and Technology of Taiwan under grant MOST 108-2634-F-002-018.  ... 
doi:10.1109/cvpr.2019.00273 dblp:conf/cvpr/ChenLLW19 fatcat:yxy47fjnnbchljkxlmge7fte3u

Survey on Semantic Stereo Matching / Semantic Depth Estimation [article]

Viny Saajan Victor, Peter Neigel
2021 arXiv   pre-print
Stereo matching is one of the widely used techniques for inferring depth from stereo images owing to its robustness and speed.  ...  Recent developments have shown that semantic cues from image segmentation can be used to improve the results of stereo matching.  ...  Joint-Learning Approach: These approaches use a common architecture for both stereo matching and semantic segmentation in the initial stages to extract features that are generic for both the tasks and  ... 
arXiv:2109.10123v1 fatcat:clqvhhh6o5bwtfvv4bir3svwqy

DTS-Net: Depth-to-Space Networks for Fast and Accurate Semantic Object Segmentation

Hatem Ibrahem, Ahmed Salem, Hyun-Soo Kang
2022 Sensors  
In addition, we explore the joint semantic segmentation and depth estimation task and demonstrate that the proposed technique can efficiently perform both tasks simultaneously, outperforming state-of-art  ...  We propose Depth-to-Space Net (DTS-Net), an effective technique for semantic segmentation using the efficient sub-pixel convolutional neural network.  ...  We also provide the test and the evaluation codes of the proposed method at: https://github.com/HatemHosam/ DTS-Net which is created and accessed on 21 June 2021.  ... 
doi:10.3390/s22010337 pmid:35009879 pmcid:PMC8749585 fatcat:mvxugub4anbmbcpoybyrmm7dxa

Auxiliary Learning for Deep Multi-task Learning [article]

Yifan Liu, Bohan Zhuang, Chunhua Shen, Hao Chen, Wei Yin
2019 arXiv   pre-print
We evaluate the proposed auxiliary module on pixel-wise prediction tasks, including semantic segmentation, depth estimation, and surface normal prediction with different network structures.  ...  Multi-task learning (MTL) is an efficient solution to solve multiple tasks simultaneously in order to get better speed and performance than handling each single-task in turn.  ...  (a) and (d): Semantic segmentation and depth estimation results of ResNet50-joint. (b) and (e): Semantic segmentation and depth estimation results of ResNet50-Auxi-all. (c) and (f): Ground truth.  ... 
arXiv:1909.02214v2 fatcat:s54gcmpghvfflhj3lmlctnukfy

Multi-task human analysis in still images: 2D/3D pose, depth map, and multi-part segmentation [article]

Daniel Sánchez, Marc Oliu, Meysam Madadi, Xavier Baró, Sergio Escalera
2019 arXiv   pre-print
Specifically, we study the correlation of 2D/3D pose estimation, body part segmentation and full-body depth estimation.  ...  the most, 2D pose improves neither 3D pose nor full-body depth estimation.  ...  In summary, ankle, wrist and elbow are the most difficult joints to learn. Again, we see those body parts and joints are difficult to predict for all tasks. D.  ... 
arXiv:1905.03003v1 fatcat:wapz62r36rhuhavfkonldyvzwq

Domain Adaptive Semantic Segmentation with Self-Supervised Depth Estimation [article]

Qin Wang, Dengxin Dai, Lukas Hoyer, Luc Van Gool, Olga Fink
2021 arXiv   pre-print
On the one hand, we propose to explicitly learn the task feature correlation to strengthen the target semantic predictions with the help of target depth estimation.  ...  Domain adaptation for semantic segmentation aims to improve the model performance in the presence of a distribution shift between source and target domain.  ...  Acknowledgement The contributions of Qin Wang and Olga Fink were funded by the Swiss National Science Foundation (SNSF) Grant no. PP00P2 176878.  ... 
arXiv:2104.13613v2 fatcat:6velaiarczhojkpyizv7app7le

Domain Adaptive Semantic Segmentation with Self-Supervised Depth Estimation

Qin Wang, Dengxin Dai, Lukas Hoyer, Luc Van Gool, Olga Fink
2021 2021 IEEE/CVF International Conference on Computer Vision (ICCV)  
On the one hand, we propose to explicitly learn the task feature correlation to strengthen the target semantic predictions with the help of target depth estimation.  ...  Domain adaptation for semantic segmentation aims to improve the model performance in the presence of a distribution shift between source and target domain.  ...  Acknowledgement The contributions of Qin Wang and Olga Fink were funded by the Swiss National Science Foundation (SNSF) Grant no. PP00P2 176878.  ... 
doi:10.1109/iccv48922.2021.00840 fatcat:2qpvrotukzfyxg3mt4unzmhufy

Guest Editors' Introduction to the Special Issue on RGB-D Vision: Methods and Applications

Mohammed Bennamoun, Yulan Guo, Federico Tombari, Kamal Youcef-Toumi, Ko Nishino
2020 IEEE Transactions on Pattern Analysis and Machine Intelligence  
The 26 accepted papers of this special issue can be grouped into six different main categories: (i) new sensing technologies, (ii) depth estimation and enhancement, (iii) simultaneous localization and  ...  Compared to 2D images and 3D data (including depth images, point clouds and meshes), RGB-D images represent both the photometric and geometric information of a scene.  ...  Yang proposes a task-recursive learning (TRL) framework to jointly and recurrently conduct three representative tasks including depth estimation, surface normal prediction and semantic segmentation.  ... 
doi:10.1109/tpami.2020.2976227 fatcat:dqt5dt3ymnesfikmgu2sxffdcu

Adversarial Attacks on Monocular Depth Estimation [article]

Ziqi Zhang, Xinge Zhu, Yingwei Li, Xiangqun Chen, Yao Guo
2020 arXiv   pre-print
Recent advances of deep learning have brought exceptional performance on many computer vision tasks such as semantic segmentation and depth estimation.  ...  We then adapt several state-of-the-art attack methods for classification on the field of depth estimation.  ...  Zhang et al proposed a task-recursive-learning framework for semantic segmentation and depth estimation [43] . B. Adversarial Attacks White-box and black-box attacks.  ... 
arXiv:2003.10315v1 fatcat:mhfqoh3dsbcvjb7csi2rdalqvu

Future Segmentation Using 3D Structure [article]

Suhani Vora, Reza Mahjourian, Soeren Pirk, Anelia Angelova
2018 arXiv   pre-print
Our framework is based on learnable sub-modules capable of predicting pixel-wise scene semantic labels, depth, and camera ego-motion of adjacent frames.  ...  Predicting the future to anticipate the outcome of events and actions is a critical attribute of autonomous agents; particularly for agents which must rely heavily on real time visual data for decision  ...  We further plan to extend this work and demonstrate its effectiveness, by predicting future events for better motion planning, e.g. in the context of human-robot interaction.  ... 
arXiv:1811.11358v1 fatcat:tbjpzays6fchznx6kekjcd3gyu
« Previous Showing results 1 — 15 out of 4,185 results