Playground (PG) for low-quality image detection. Images of PG are shot by two different views of cameras, pedestrian images are taken at different periods, including day and night. There are a total of 5,752 labeled images with 31,041 annotations. The average size of the pedestrian is 41*87 and the image size is 480x640.
-Baidu(Link：https://pan.baidu.com/s/1oJMp3SQLtJ6yv1bVZfff1w Extraction code:u2mn )
-google(https://drive.google.com/drive/folders/12Bn4FExI-tl6_LjXY-OzPHJsYf8AURrI?usp=sharing)
-Baidu(Link：链接：https://pan.baidu.com/s/1d5hM4YowTSvdylZuqHAswg 提取码：qscq )
-google(Link：https://drive.google.com/file/d/1eUrr4dLgq8BQcf2pzZ8bSU8hKFfDAkA5/view?usp=sharing)
The labeling format is .xml format, which is the same as the PASCAL VOC data set. It can be easily trained and tested on common target detection algorithms (i.e. Faster R-CNN, YOLOv3, YOLOv4, etc.).
If you use this method or this code in your research, please cite as:
@article{jin2021pedestrian,
title={Pedestrian detection with super-resolution reconstruction for low-quality image},
author={Jin, Yi and Zhang, Yue and Cen, Yigang and Li, Yidong and Mladenovic, Vladimir and Voronin, Viacheslav},
journal={Pattern Recognition},
volume={115},
pages={107846},
year={2021},
publisher={Elsevier}
}
This code is released for academic research / non-commercial use only. If you wish to use for commercial purposes, please contact Yue Zhang by email [email protected].