Improving Object Detection to Fisheye Cameras with Open-Vocabulary Pseudo-Label Approach

Long Hoang Pham, Quoc Pham-Nam Ho, Duong Nguyen-Ngoc Tran, Tai Huu-Phuong Tran, Huy-Hung Nguyen, Duong Khac Vu, Chi Dai Tran, Ngoc Doan-Minh Huynh, Hyung-Min Jeon, Hyung-Joon Jeon, Jae Wook Jeon; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2024, pp. 7100-7107

Fish-eye cameras have long been employed in traffic surveillance systems to allow for wider observation of the roads. Despite their widespread use limited computer vision research is tailored explicitly to images captured by fish-eye cameras. The AI City Challenge 2024 - Track 4 introduces a novel fish-eye camera dataset for the 2D road object detection task. This paper proposes a framework designed to detect objects in fish-eye camera images. Our approach involves several key steps: first we generate im- age data to bridge the representation gap between day and night images. Next we leverage zero-shot open vocabulary detection to produce pseudo-labels aiding in training supervised object detection models. Additionally we optimize the model's hyper-parameters and inference configuration for better performance. Finally we apply various post-processing techniques to enhance detection performance. Our solution achieves a final F1 score of 0.6194 in the AI City Challenge 2024 - Track 4 ranking third among competing teams. The source code is available at GitHub Repo.