Pedestrian detection from fisheye videos with robustness to occlusion

CONTEXT This internship is part of the development of an autonomous drone equipped with a fisheye camera for intelligent surveillance in urban environments. Fisheye cameras offer a very wide field of vision (>180°) but introduce significant geometric distortions. The goal is to develop a robust pedestrian detection method for fisheye images that can handle occlusions. INTERNSHIP OBJECTIVES • Accurate detection of pedestrians in fisheye videos • Robustness to partial and total occlusions • Integration with a real-time detector (YOLO) TASKS REQUIRED • Bibliographic study on resolving occlusions in pedestrian detection • Development of a spatio-temporal attention module • Implementation in Python/PyTorch • Integration with YOLO for real-time detection • Comparison with the state of the art • Writing of a scientific article DELIVERABLES • Spatio-temporal attention module for occlusion resolution • Validated integration with YOLO • Comparative benchmark with the state of the art • Scientific article