相关参考文献
注意：仅列出部分参考文献，下载原文获取全部文献信息。
Article
Computer Science, Artificial Intelligence
Xiaowei Yang et al.
Summary: The research proposes RDNet, which incorporates edge cues into stereo matching and generates a depth ground-truth boundary dataset by mining instance segmentation and semantic segmentation datasets. Additionally, methods like multi-scale cost volume and disparity refinement network are introduced to further optimize stereo matching performance.
IMAGE AND VISION COMPUTING
(2022)
Article
Robotics
Wei Chen et al.
Summary: The dimensions of cost volumes and aggregation networks are crucial for balancing the speed and accuracy of stereo matching. In order to achieve a balance between speed and precision, a novel stereo framework combining 2D and 3D networks is proposed. By cooperating with each other, the framework enables fast and accurate stereo matching results.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Gangwei Xu et al.
Summary: This paper proposes a novel method for constructing cost volume, which generates attention weights to suppress redundant information and enhance matching-related information. By introducing multi-level adaptive patch matching, the distinctiveness of the matching cost at different disparities is improved. The proposed attention concatenation volume (ACV) can be seamlessly embedded into stereo matching networks, achieving higher accuracy with a more lightweight aggregation network. The highly accurate network (ACVNet) based on ACV achieves state-of-the-art performance on several benchmarks.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Article
Computer Science, Artificial Intelligence
Wanshui Gan et al.
Summary: The lightweight adaptive network (LWANet) proposed in this paper combines self-supervised learning methods to achieve online adaptive stereo depth estimation with low computation cost and GPU memory space. By utilizing pseudo 3D convolution and CSPN network, the network strikes a better balance between accuracy and computational cost, effectively addressing the domain shift problem for embedded devices like NVIDIA Jetson TX2.
Proceedings Paper
Automation & Control Systems
Xiaogang Jia et al.
Summary: MSCVNet combines traditional methods and neural networks to improve the quality of the cost volume, achieving a balance between speed and accuracy. It also includes a specially designed algorithm to handle loss in discontinuous areas, leading to a significant improvement in accuracy compared to other methods.
2021 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA 2021)
(2021)
Proceedings Paper
Automation & Control Systems
Antyanta Bangunharcana et al.
Summary: This study introduces a volumetric deep learning approach for stereo matching, using guided cost volume excitation and top-k selection to optimize disparity estimation, resulting in an end-to-end network called CoEx. Experimental results demonstrate the efficiency and performance of the model across various datasets, surpassing other speed-based algorithms and remaining competitive with state-of-the-art algorithms.
2021 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS)
(2021)
Proceedings Paper
Computer Science, Artificial Intelligence
Bin Xu et al.
Summary: This paper introduces a novel module for stereo matching networks that improves real-time performance and maintains accuracy. The module can be seamlessly embedded into existing networks to achieve significant acceleration. The real-time network based on this module outperforms existing networks on the KITTI dataset.
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021
(2021)
Article
Computer Science, Hardware & Architecture
Yun Xie et al.
Summary: In this article, an end-to-end real-time stereo matching network (RTSMNet) is proposed, consisting of three modules: global and local feature extraction, initial disparity estimation, and feature-guided spatial attention upsampling. The network achieves high frame rates and competitive accuracy on various devices, which is significant for the field of image processing.
Article
Computer Science, Artificial Intelligence
Zhengfa Liang et al.
Summary: In this paper, an end-to-end trainable convolutional neural network is proposed for stereo matching, with shared feature extraction, initial disparity estimation, and disparity refinement as its three sub-modules, achieving good performance across multiple datasets.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Xiao Song et al.
INTERNATIONAL JOURNAL OF COMPUTER VISION
(2020)
Article
Computer Science, Artificial Intelligence
Xinjing Cheng et al.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2020)
Article
Computer Science, Information Systems
Xiaowei Yang et al.
Article
Engineering, Electrical & Electronic
Zhibo Rao et al.
APSIPA TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING
(2020)
Proceedings Paper
Computer Science, Artificial Intelligence
Alessio Tonioni et al.
2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019)
(2019)
Proceedings Paper
Computer Science, Artificial Intelligence
Alex Kendall et al.
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Clement Godard et al.
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Francois Chollet
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017)
(2017)
Proceedings Paper
Computer Science, Artificial Intelligence
Zhuoyuan Chen et al.
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2015)
Proceedings Paper
Computer Science, Artificial Intelligence
Chenyi Chen et al.
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2015)
Article
Engineering, Industrial
Jian Hou et al.
INDUSTRIAL ROBOT-THE INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH AND APPLICATION
(2007)
