Beyond Overall Accuracy: Pose- and Occlusion-driven Fairness Analysis in Pedestrian Detection for Autonomous Driving
Mohammad Khoshkdahan, Arman Akbari, Arash Akbari, Xuan Zhang · 2025
Read Beyond Overall Accuracy: Pose- and Occlusion-driven Fairness Analysis in Pedestrian Detection for Autonomous Driving by Mohammad Khoshkdahan, Arman Akbari, Arash Akbari, Xuan Zhang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-26166.txt.
Description
Khoshkdahan and three co-authors check whether pedestrian detectors for self-driving cars treat all walkers fairly. A detector can post a fine average score while missing some poses far more often. The team tested eight models on a dataset rich in body-pose labels.
They measured fairness with a gap score across leg and elbow positions, body orientation, and hidden joints. Detectors struggled most with parallel legs, straight elbows, and side views. Blocked lower-body joints hurt detection more than blocked heads. One model, Cascade R-CNN, missed least overall and showed the smallest bias. The authors call for fairness checks beside accuracy in safety testing.
Catalog record
| CREATOR | Mohammad Khoshkdahan, Arman Akbari, Arash Akbari, Xuan Zhang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-26166