An efficient algorithm for pothole detection using stereo vision

Abstract

The advent of learning with noisy labels (LNL), multi-rater learning, and human-AI collaboration has revolutionised the development of robust classifiers, enabling them to address the challenges posed by different types of data imperfections and complex decision processes commonly encountered in real-world applications. While each of these methodologies has individually made significant strides in addressing their unique challenges, the development of techniques that can simultaneously tackle these three problems remains underexplored. This paper addresses this research gap by integrating noisy-label learning, multi-rater learning, and human-AI collaboration with new benchmarks and the innovative Learning to Complement with Multiple Humans (LECOMH) approach. LECOMH optimises the level of human collaboration during testing, aiming to optimise classification accuracy while minimising collaboration costs that vary from 0 to M , where M is the maximum number of human collaborators. We quantitatively compare LECOMH with leading human-AI collaboration methods using our proposed benchmarks. LECOMH consistently outperforms the competition, with accuracy improving as collaboration costs increase. Notably, LECOMH is the only method enhancing human labeller performance across all benchmarks.

Pothole Detection Based on Disparity Transformation and Road Surface Modeling

“…15, and the corresponding ground truth is shown in the sixth row. We also compare our proposed algorithm with those produced in [15] and [16]. The pothole detection results obtained using the algorithms presented in [15] and [16] are shown in the third and forth rows of Fig.…”

“…We also compare the proposed algorithm with [15] and [16] with respect to the pixel-level precision, recall, F-score and accuracy: F-score = 2 × precision × recall precision + recall ,…”

“…For example, Zhang and Elaksher [13] mounted a single camera on an unmanned aerial vehicle (UAV) to reconstruct the road surface via Structure from Motion (SfM) [18]. A variety of stereo vision-based pothole detection methods have been developed as well [15], [16]. The 3D point cloud generated from a disparity map was interpolated into a quadratic surface using least squares fitting (LSF) [15].…”

“…A variety of stereo vision-based pothole detection methods have been developed as well [15], [16]. The 3D point cloud generated from a disparity map was interpolated into a quadratic surface using least squares fitting (LSF) [15]. The potholes were then recognized by comparing the difference between the 3D point cloud and the fitted quadratic surface.…”

“…Experimental results of pothole detection: (a) left stereo images; (b) transformed disparity maps; (c) pothole detection results obtained using the algorithm proposed in[15]; (d) pothole detection results obtained using the algorithm presented in[16]; (e) pothole detection results obtained using the proposed algorithm; (f) pothole ground truth.…”