RoboCup Soccer Dataset
收藏资源简介:
该数据集名为RoboCup Soccer Dataset,由罗马第一大学的研究团队创建,包含10,000张来自户外RoboCup SPL比赛的标记图像。该数据集用于验证一种新的自监督学习框架,该框架通过生成伪标签来增强足球机器人上的球检测性能。数据集的创建过程包括从NAO机器人日志中检索数据,使用YOLO-World模型进行预标注,然后进行手动细化。数据集旨在解决在动态和具有挑战性的环境中进行球检测的问题,并已公开发布供社区使用。
This dataset, named RoboCup Soccer Dataset, was created by a research team from Sapienza University of Rome, and contains 10,000 annotated images from outdoor RoboCup SPL matches. It is utilized to validate a novel self-supervised learning framework that enhances ball detection performance on soccer robots by generating pseudo-labels. The dataset creation process involves retrieving data from NAO robot logs, performing pre-annotation with the YOLO-World model, followed by manual refinement. This dataset aims to address the challenge of ball detection in dynamic and demanding environments, and has been publicly released for use by the global research community.

- 1Self-supervised Feature Extraction for Enhanced Ball Detection on Soccer Robots罗马第一大学 · 2025年



