InfiniteForm
收藏资源简介:
InfiniteForm是由边缘分析公司创建的一个开源合成数据集,专门针对健身领域。该数据集包含60000张图像,涵盖15种健身姿势类别,包括单人和多人场景,以及真实的光照、摄像机角度和遮挡变化。数据集设计旨在最小化性别、肤色和体型偏见,提供包括2D关键点、深度和遮挡在内的精确标签。创建过程中采用了Blender和物理基础渲染引擎,以及SMPL-X身体模型,确保了图像的真实性和多样性。InfiniteForm数据集主要用于训练和评估计算机视觉模型,以解决远程健身中的视觉问题,如姿势矫正、能量消耗估计和重复计数等。
InfiniteForm is an open-source synthetic dataset created by Edge Analytics, specifically targeting the fitness domain. This dataset contains 60,000 images covering 15 fitness pose categories, including single-person and multi-person scenarios, as well as variations in real-world lighting, camera angles, and occlusions. The dataset is designed to minimize biases related to gender, skin color, and body type, and provides precise annotations including 2D keypoints, depth information, and occlusion status. Blender, physically based rendering engines, and the SMPL-X body model were utilized during the dataset creation process to ensure the authenticity and diversity of the images. The InfiniteForm dataset is primarily used for training and evaluating computer vision models to address visual challenges in remote fitness scenarios, such as pose correction, energy expenditure estimation, and repetition counting.

- 1InfiniteForm: A synthetic, minimal bias dataset for fitness applications边缘分析公司 · 2021年



