MVA2023 鸟类侦测小目标检测挑战赛：数据集、方法与结果
计算机视觉与模式识别
2023-09-14 v1 机器学习
摘要
小目标检测（SOD）是一个重要的机器视觉课题，因为（i）多种现实应用需要对远距离目标进行检测，且（ii）SOD 因小目标图像表现噪声大、模糊且信息量少而是一项具挑战性的任务。本文提出一个包含 39,070 张图像、137,121 个鸟类实例的新 SOD 数据集，称为面向鸟类侦测的小目标检测（SOD4SB）数据集。本文介绍了基于 SOD4SB 数据集的挑战赛详情。总计 223 名参与者加入该挑战赛。本文简要介绍获奖方法。该数据集、基线代码及用于公开测试集评估的网站均已公开可用。
关键词
引用
@article{arxiv.2307.09143,
title = {MVA2023 Small Object Detection Challenge for Spotting Birds: Dataset, Methods, and Results},
author = {Yuki Kondo and Norimichi Ukita and Takayuki Yamaguchi and Hao-Yu Hou and Mu-Yi Shen and Chia-Chi Hsu and En-Ming Huang and Yu-Chen Huang and Yu-Cheng Xia and Chien-Yao Wang and Chun-Yi Lee and Da Huo and Marc A. Kastner and Tingwei Liu and Yasutomo Kawanishi and Takatsugu Hirayama and Takahiro Komamizu and Ichiro Ide and Yosuke Shinya and Xinyao Liu and Guang Liang and Syusuke Yasui},
journal= {arXiv preprint arXiv:2307.09143},
year = {2023}
}
备注
This paper is included in the proceedings of the 18th International Conference on Machine Vision Applications (MVA2023). It will be officially published at a later date. Project page : https://www.mva-org.jp/mva2023/challenge