Title: Deep learning approaches for pedestrian detection for enhancing safety of autonomous vehicles

Speaker: Mr. Muhammad Mobaidul Islam

Date: Friday, October 28, 2022

Pedestrian detection is an essential task in autonomous driving but often is challenged due to partial occlusion, different pose and body articulation of pedestrians, and sensitivity of sensors under ambient lighting conditions. Among different approaches, network fusion has been recently explored as an approach for improving pedestrian detection performance. However, most existing fusion methods suffer from runtime efficiency, modularity, scalability, and maintainability due to the complex structure of the entire fused models, their end-to-end training requirements, and sequential fusion process. Addressing these challenges, here we discuss a novel fusion framework that combines asymmetric inferences from object detectors and semantic segmentation networks for jointly detecting multiple pedestrians. This is achieved by introducing a consensus-based scoring method that fuses pair-wise pixel-relevant information from the object detector and the semantic segmentation network to boost the final confidence scores. The parallel implementation of the object detection and semantic segmentation networks in the fusion framework entails a low runtime overhead. The efficiency and robustness of the fusion framework are extensively evaluated by fusing different state-of-the-art pedestrian detectors and semantic segmentation networks on a public dataset. The generalization of fused models is also examined on new cross pedestrian data collected through an autonomous car. Results show that the fusion method significantly improves detection performance while achieving competitive runtime efficiency.