Awesome

FloW-Dataset

</div> </a>

Introduction

Marine wastes are severely threatening marine animals and their habitat, also causing an impact on human life through toxic substances transportation and accumulation. To prevent the wastes especially the plastic trash from getting into the ocean, it is essential to detect and clean the floating wastes in inland waters efficiently like in rivers, lakes, and canals.

FloW is the first dataset for floating waste detection in inland waters. It contains a vision-based sub-dataset, FloW-Img, and a multimodal dataset, FloW-RI which contains the spatial and temporal calibrated image and millimeter-wave radar data.

By publishing Flow, it is hoped that more attention from research communities could be paid to floating waste detection in inland waters as well as the challenging small object detection over the water surface. In addition, waste detection based on millimeter-wave radar data or the fusion of image and radar data is also a novel task and FloW provides accessible real-world data.

</div> </a>

Dataset Download

Our paper has been accepted by ICCV2021. To know more details about FloW dataset, our paper can be found here. If you find our dataset useful for your research, you can cite:

@InProceedings{Cheng_2021_ICCV, 
  author = {Cheng, Yuwei and Zhu, Jiannan and Jiang, Mengxin and Fu, Jie and Pang, Changsong and Wang, Peidong and Sankaran, Kris and Onabola, Olawale and Liu, Yimin and Liu, Dianbo and Bengio, Yoshua}, 
  title = {FloW: A Dataset and Benchmark for Floating Waste Detection in Inland Waters}, 
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)}, 
  month = {October}, 
  year = {2021}, 
  pages = {10953-10962}
}

The dataset has been published now. To learn more details about FloW and download the dataset with the development tools, you can visit our website (Chinese or English), and apply for downloading the dataset in "Customer Support -> Dataset -> FloW". Let’s work on floating waste detection together!