首届微小目标检测挑战赛：方法与结果
计算机视觉与模式识别
2020-10-07 v2
摘要
首届微小目标检测（TOD）挑战赛旨在鼓励研究开发新颖且准确的图像微小目标检测方法，这类图像具有广阔视野，当前聚焦于微小行人检测。TinyPerson 数据集用于 TOD 挑战赛并已公开发布。它包含 1610 张图像和 72651 个框级标注。全球约 36 支参赛队伍参加了首届 TOD 挑战赛。本文对首届 TOD 挑战赛进行简要总结，包括前三名方法的简介。提交排行榜将重新开放给对 TOD 挑战赛感兴趣的研究人员。基准数据集及其他信息可在 https://github.com/ucas-vg/TinyBenchmark 获取。
引用
@article{arxiv.2009.07506,
title = {The 1st Tiny Object Detection Challenge:Methods and Results},
author = {Xuehui Yu and Zhenjun Han and Yuqi Gong and Nan Jiang and Jian Zhao and Qixiang Ye and Jie Chen and Yuan Feng and Bin Zhang and Xiaodi Wang and Ying Xin and Jingwei Liu and Mingyuan Mao and Sheng Xu and Baochang Zhang and Shumin Han and Cheng Gao and Wei Tang and Lizuo Jin and Mingbo Hong and Yuchao Yang and Shuiwang Li and Huan Luo and Qijun Zhao and Humphrey Shi},
journal= {arXiv preprint arXiv:2009.07506},
year = {2020}
}
备注
ECCV2020 Workshop on Real-world Computer Vision from Inputs with Limited Quality (RLQ) and Tiny Object Detection Challenge