PPD: A New Valet Parking Pedestrian Fisheye Dataset for Autonomous Driving

• DOI: 10.48550/arxiv.2309.11002
• Publication Date: 2023-01-01
• Authors: Zizhang Wu; Xinyuan Chen; Fan Song; Yuanzhu Gan; Tianhao Xu; Jian Pu; Rui Tang
• Keywords: Baseline (sea); Pedestrian detection

Pedestrian detection under valet parking scenarios is fundamental for autonomous driving. However, the presence of pedestrians can be manifested in a variety ways and postures imperfect ambient conditions, which adversely affect performance. Furthermore, models trained on publicdatasets that include generally provide suboptimal outcomes these scenarios. In this paper, wepresent Parking Dataset (PPD), large-scale fisheye dataset to support research dealing with real-world pedestrians, especially occlusions diverse postures. PPD consists several distinctive types captured cameras. Additionally, we present pedestrian baseline dataset, introduce two data augmentation techniques improve by enhancing diversity ofthe original dataset. Extensive experiments validate effectiveness our novel approaches over baselinesand dataset's exceptional generalizability.