SDR-GAIN: A High Real-Time Occluded Pedestrian Pose Completion Method for Autonomous Driving
This addresses a critical safety issue for autonomous driving systems by improving pose estimation under occlusion, though it is incremental as it builds on prior imputation techniques.
The paper tackled the problem of accurately reconstructing occluded pedestrian keypoints in autonomous driving by proposing SDR-GAIN, a real-time framework that learns from keypoint coordinate distributions to interpolate missing positions, achieving microsecond-level inference and outperforming existing methods on COCO and JAAD datasets.
With the advancement of vision-based autonomous driving technology, pedestrian detection have become an important component for improving traffic safety and driving system robustness. Nevertheless, in complex traffic scenarios, conventional pose estimation approaches frequently fail to accurately reconstruct occluded keypoints, primarily due to obstructions caused by vehicles, vegetation, or architectural elements. To address this issue, we propose a novel real-time occluded pedestrian pose completion framework termed Separation and Dimensionality Reduction-based Generative Adversarial Imputation Nets (SDR-GAIN). Unlike previous approaches that train visual models to distinguish occlusion patterns, SDR-GAIN aims to learn human pose directly from the numerical distribution of keypoint coordinates and interpolate missing positions. It employs a self-supervised adversarial learning paradigm to train lightweight generators with residual structures for the imputation of missing pose keypoints. Additionally, it integrates multiple pose standardization techniques to alleviate the difficulty of the learning process. Experiments conducted on the COCO and JAAD datasets demonstrate that SDR-GAIN surpasses conventional machine learning and Transformer-based missing data interpolation algorithms in accurately recovering occluded pedestrian keypoints, while simultaneously achieving microsecond-level real-time inference.