深度学习论文: Computer Vision for Road Imaging and Pothole Detection: A State-of-the-Art Review of Systems and Algorithms
Computer Vision for Road Imaging and Pothole Detection: A State-of-the-Art Review of Systems and Algorithms
PDF: https://arxiv.org/pdf/2204.13590.pdf PyTorch代码: https://github.com/shanglianlm0525/CvPytorch PyTorch代码: https://github.com/shanglianlm0525/PyTorch-Networks
1 概述
这篇文章详述了道路成像的传感器和用于坑洞检测的相关算法以及开源的数据集。
2 Road Imaging Systems
常用的道路成像系统包括：Laser scanners， Microsoft Kinect sensors 和 Stereo Camera(s)。
此外，还有基于多视角几何原理的单目运动相机或者多目相机。
3 Road Pothole Detection Approaches
3-1 Classical 2-D Image Processing
传统的2D图像处理主要涉及四个阶段：(1) image pre-processing, (2) image segmentation, (3) damaged area extraction, and (4) detection result post-processing。
3-2 3-D Point Cloud Modeling and Segmentation
3-D road point clouds的方法主要分为两个阶段：(1) interpolating the observed 3-D road point cloud into an explicit geometric model (typically a planar or quadratic surface), and (2) segmenting the observed 3-D road point cloud by comparing it with the interpolated geometric model.
3-3 Machine/Deep Learning
3-3-1 Image Classification-Based Methods
3-3-2 Object Detection-Based Methods
3-3-3 Semantic Segmentation-Based Methods
3-4 Hybrid Methods
Hybrid 通常使用两种或以上的算法来发现地坑缺陷。
4 Public Datasets
1 road image classification
2 instance-level pothole detection
3 Indian roads with semantic segmentation
4 CIMAT Challenging Sequences for Autonomous Driving (CCSAD)
5 Japan road damage dataset
6 automatic pothole detection and localization in urban streets
7 binary road image classification
8 multi-modal road pothole detection dataset
9 Pothole-600
