Recent methods in stereo matching have continuously improved the accuracy using deep models. This gain, however, is attained with a high increase in computation cost, such that the network may not fit even on a moderate GPU. This issue raises problems when the model needs to be deployed on resource-limited devices. For this, we propose two light models for stereo vision with reduced complexity and without sacrificing accuracy. Depending on the dimension of cost volume, we design a 2D and a 3D model with encoder-decoders built from 2D and 3D convolutions, respectively. To this end, we leverage 2D MobileNet blocks and extend them to 3D for stereo vision application. Besides, a new cost volume is proposed to boost the accuracy of the 2D model, making it performing close to 3D networks. Experiments show that the proposed 2D/3D networks effectively reduce the computational expense (27%/95% and 72%/38% fewer parameters/operations in 2D and 3D models, respectively) while upholding the accuracy. Our code is available at https://github.com/cogsys-tuebingen/mobilestereonet.
Lightweight Multiplane Images Network for Real-Time Stereoscopic Conversion from Planar Video
Hefei University of Technology（合肥工业大学）；Peng Cheng National Laboratory（鹏城国家实验室）；Anhui University（安徽大学）；Peking University Shenzhen Graduate School（北京大学深圳研究生院）
A Lightweight Target-Driven Network of Stereo Matching for Inland Waterways
College of Artificial Intelligence, Tianjin University of Science and Technology（天津科技大学人工智能学院）；School of Artificial Intelligence, Anhui University（安徽大学人工智能学院）；Tianjin Institute of Navigation Instruments（天津航海仪器研究所）
Lightweight and Accurate Multi-View Stereo with Confidence-Aware Diffusion Model
Department of Computer Science, ETH Zurich（苏黎世联邦理工学院计算机科学系）；College of Computing and Data Science, Nanyang Technological University（南洋理工大学计算机科学与数据科学学院）；Center for Frontier AI Research, Institute of High Performance Computing, A*STAR（A*STAR前沿人工智能研究中心，高性能计算研究所）；Microsoft, Zurich（微软（瑞士））
BDIS-SLAM: A lightweight CPU-based dense stereo SLAM for surgery
United Imaging Research Institute of Intelligent Imaging（联影智能影像研究院）；University of Michigan（密歇根大学）；University of Technology Sydney（悉尼科技大学）；Imperial College London（伦敦帝国理工学院）
StereoVAE: A lightweight stereo-matching system using embedded GPUs
School of Computing, Tokyo Institute of Technology（东京技术大学计算机学系）；School of Electronic Science & Engineering, Nanjing University（南京大学电子科学与工程学院）；Artificial Intelligence Research Center, National Institute of Advanced Industrial Science and Technology (AIST)（国家先进工业科学与技术研究院人工智能研究中心）
Distill-then-prune: An Efficient Compression Framework for Real-time Stereo Matching Network on Edge Devices
UBTech Robotics Corp（优必选机器人公司）；Shenzhen Institute of Advanced Technology（深圳先进技术研究院）；Chinese Academy of Sciences（中国科学院）；Beijing University of Posts and Telecommunications（北京邮电大学）
Learning Inverse Depth Regression for Multi-View Stereo with Correlation Cost Volume
Huazhong University of Science and Technology（华中科技大学）
深度学习已被证明对于多视图立体视觉（MVS）中的深度推断是有效的。然而，在这个领域的可扩展性和准确性仍然是一个悬而未决的问题。这可以归因于内存消耗成本体积表示和不适当的深度推断。受立体匹配中组间相关性的启发，我们提出了一种平均组间相关性相似性度量来构造一个轻量级的代价体积。这不仅可以减少内存消耗，而且还可以减少成本卷过滤中的计算负担。基于我们的有效成本体积表示，我们提出了一个级联的3D U-网络模块，以规范的成本体积，以进一步提高性能。与将多视图深度推断处理为深度回归问题或逆深度分类问题的先前方法不同，我们将多视图深度推断重铸为逆深度回归任务。这使得我们的网络能够实现亚像素估计，并适用于大规模场景。通过对DTU数据集和Tanks and Temples数据集的广泛实验，我们表明我们提出的具有相关成本体积和逆深度回归（CIDER）的网络实现了最先进的结果，证明了其在可扩展性和准确性方面的优越性能。
本文提出了一种基于长短期记忆（LSTM）的自适应聚合递归多视图立体视觉网络，即AA-RMVSNet。我们首先引入了一个视图内聚合模块，通过使用上下文感知卷积和多尺度聚合自适应地提取图像特征，有效地提高了具有挑战性的区域，如薄的物体和大的低纹理表面的性能。为了克服复杂场景中变化遮挡的困难，我们提出了一个视图间成本体积聚合模块，用于自适应像素视图聚合，该模块能够在所有视图之间保持更好的匹配对。提出的两个自适应聚合模块是轻量级的，有效的和互补的提高三维重建的准确性和完整性。代替传统的3D CNN，我们利用具有递归结构的混合网络进行成本体积正则化，这允许高分辨率重建和更精细的假设平面扫描。所提出的网络是端到端训练的，在各种数据集上都取得了优异的性能。该算法在Tanks and Temples基准测试中对$1^{st}$进行了排序，并在DTU数据集上取得了较好的结果，具有较强的泛化能力和鲁棒性。我们的方法的实现可在https://github.com/QT-Zhu/AA-RMVSNet上获得。
SNNSIR: A Simple Spiking Neural Network for Stereo Image Restoration
School of Big Data and Software Engineering, Chongqing University（大数据与软件工程学院，重庆大学）；Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）；School of Microelectronics and Communication Engineering, Chongqing University（微电子与通信工程学院，重庆大学）；School of Electrical and Information Engineering, Tianjin University（电气与信息工程学院，天津大学）
ESMStereo: Enhanced ShuffleMixer Disparity Upsampling for Real-Time and Accurate Stereo Matching
Center for Mathematical Modelling and Intelligent Systems for Health and Environment (MISHE)（数学建模与智能系统用于健康和环境的中心（MISHE））；Atlantic Technological University（大西洋技术大学）；Department of Engineering and Physics（工程与物理系）
Training Deep Stereo Matching Networks on Tree Branch Imagery: A Benchmark Study for Real-Time UAV Forestry Applications
Centre for Data Science and Artificial Intelligence（数据科学与人工智能中心）；Victoria University of Wellington（威灵顿维多利亚大学）；Department of Computer Science and Software Engineering（计算机科学与软件工程系）；University of Canterbury（坎特伯雷大学）
FFCA-Net: Stereo Image Compression via Fast Cascade Alignment of Side Information
Tsinghua Shenzhen International Graduate School, Tsinghua University（清华大学深圳国际研究生院）；Research Center of Artificial Intelligence, Peng Cheng Laboratory（鹏城实验室人工智能研究中心）
LightStereo: Channel Boost Is All You Need for Efficient 2D Cost Aggregation
School of Computer Science, Wuhan University（武汉大学计算机学院）；Waytous（万图思睿）；University of Bologna（博洛尼亚大学）；Rock Universe AI；University of California, Berkeley（加州大学伯克利分校）；Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；Metoak（米托克）
Learning Robust Stereo Matching in the Wild with Selective Mixture-of-Experts
City University of Hong Kong（香港城市大学）；Shenzhen Campus, Sun Yat-sen University（中山大学深圳校区）；Chinese Academy of Sciences（中国科学院）；Zhejiang University（浙江大学）；South China Normal University（华南师范大学）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））
