The Way Up: A Dataset for Hold Usage Detection in Sport Climbing
This work addresses a problem for researchers and developers in sports climbing AI by providing the first detailed dataset, though it is incremental as it applies existing pose-estimation methods to a new domain.
The authors tackled the lack of annotated data for hold usage detection in sport climbing by introducing a dataset of 22 climbing videos with ground-truth labels for hold locations, usage order, and time, and evaluated keypoint-based 2D pose-estimation models on it, achieving results that highlight climbing-specific challenges.
Detecting an athlete's position on a route and identifying hold usage are crucial in various climbing-related applications. However, no climbing dataset with detailed hold usage annotations exists to our knowledge. To address this issue, we introduce a dataset of 22 annotated climbing videos, providing ground-truth labels for hold locations, usage order, and time of use. Furthermore, we explore the application of keypoint-based 2D pose-estimation models for detecting hold usage in sport climbing. We determine usage by analyzing the key points of certain joints and the corresponding overlap with climbing holds. We evaluate multiple state-of-the-art models and analyze their accuracy on our dataset, identifying and highlighting climbing-specific challenges. Our dataset and results highlight key challenges in climbing-specific pose estimation and establish a foundation for future research toward AI-assisted systems for sports climbing.