The ability to accurately detect pedestrians in high-density urban environments is essential for ensuring the safety and efficiency of autonomous vehicles and advanced driver-assistance systems (ADAS). Traditional pedestrian detection methods face significant challenges, such as partial occlusions, variations in pedestrian appearance, and the dynamic nature of crowded scenes. This work presents an innovative algorithm for in-vehicle cameras, designed specifically to overcome these challenges by leveraging advancements in computer vision and deep learning. The proposed method integrates a feature extraction framework with a multi-scale detection strategy to enhance the identification of pedestrians in diverse and complex scenarios. By employing real-time data augmentation and adaptive thresholding techniques, the algorithm demonstrates resilience to varying lighting conditions and rapid scene changes. Extensive experiments conducted on publicly available pedestrian detection datasets confirm the effectiveness of the proposed method, showing significant improvements in precision, recall, and processing speed over existing approaches. Furthermore, the algorithm’s low computational overhead makes it well-suited for deployment in real-world vehicular systems. The proposed system not only advances the state-of-the-art in high-density pedestrian detection but also contributes to the broader goal of creating safer and more reliable autonomous driving systems. Future extensions of this work aim to incorporate predictive models for pedestrian behavior to further enhance system robustness.
Advanced In-Vehicle Camera Algorithm for High-Density Pedestrian Detection
28.04.2025
432693 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Off-Vehicle Evaluation of Camera-Based Pedestrian Detection
British Library Conference Proceedings | 2012
|Camera Based Pedestrian Detection
Springer Verlag | 2015
|