The Way Up
收藏资源简介:
The Way Up 数据集包含了22个标注的攀岩视频,提供了用于攀岩中握点使用检测的地面真实标签,包括握点位置、使用顺序和使用时间。数据集旨在解决攀岩中运动员位置检测和握点使用识别的问题,这对于自动化比赛评分、运动和动作分析等攀岩相关应用至关重要。数据集使用基于关键点的2D姿态估计模型来检测攀岩中的握点使用,并评估了多个最先进的模型在其上的准确性,突出了攀岩特有的挑战。数据集和应用结果为攀岩特定姿态估计和未来研究奠定了基础,以开发实时、便携的攀岩辅助工具,特别是为视力障碍的攀岩者。
The Way Up Dataset contains 22 annotated climbing videos, providing ground-truth labels for handhold usage detection in climbing, including handhold positions, usage sequences, and usage durations. This dataset aims to address the challenges of climber position detection and handhold usage recognition in climbing, which is critical for climbing-related applications such as automated competition scoring, sports and motion analysis. The dataset employs keypoint-based 2D pose estimation models for handhold usage detection in climbing, and evaluates the accuracy of multiple state-of-the-art models on it, highlighting the challenges unique to climbing. The dataset and its application results lay a foundation for climbing-specific pose estimation and future research aimed at developing real-time, portable climbing assistance tools, particularly for visually impaired climbers.

- 1The Way Up: A Dataset for Hold Usage Detection in Sport Climbing奥地利上奥地利应用科学大学 · 2025年



