Voxel51/fisheye8k

FishEye8K是一个用于鱼眼相机物体检测的基准数据集，包含8000张图像和157K个边界框，覆盖五个类别（行人、自行车、汽车、公交车和卡车）。数据集由18个鱼眼相机在台湾新竹的交通监控中录制，分辨率为1080×1080和1280×1280。数据集由Munkhjargal Gochoo等人策划，并由Harpreet Sahota共享，使用Creative Commons Attribution-NonCommercial-ShareAlike 4.0许可。数据集的创建过程包括数据采集、标注规则、标注过程和数据清理。为了保护隐私，数据集中的行人和车辆车牌信息进行了模糊处理。

The FishEye8K dataset is a dataset for object detection tasks using fisheye cameras, containing 8,000 images with 157,012 bounding boxes across five classes (Pedestrian, Bike, Car, Bus, and Truck). These images were recorded in 22 videos using 18 fisheye cameras in Hsinchu, Taiwan, at resolutions of 1080×1080 and 1280×1280. The dataset was curated by Munkhjargal Gochoo et al. and is shared by Harpreet Sahota. The dataset is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0. The README file also includes information about the datasets structure, creation process, and anonymization measures taken to protect privacy. The citation for the dataset is also provided.