ExoNet Database: Wearable Camera Images of Human Locomotion Environments

Abstract: Recent advances in computer vision and artificial intelligence have allowed researchers to develop environment recognition systems for lower-limb exoskeletons and prostheses. However, small-scale and private training datasets have impeded the widespread development and dissemination of image classification algorithms for human locomotion environment recognition. To address these shortcomings, we developed “ExoNet” - the first open-source, large-scale hierarchical database of high-resolution wearable camera images of human locomotion environments. Unparalleled in scale and diversity, ExoNet comprises over 5.6 million images of different indoor and outdoor real-world walking environments, which were collected using a lightweight wearable smartphone camera system throughout the summer, fall, and winter seasons. Approximately 940,000 images in ExoNet were human-annotated using a 12-class hierarchical labelling architecture. Available publicly through the IEEE DataPort repository, ExoNet offers an unprecedented community-based platform for training, developing, and comparing next-generation image classification algorithms for human locomotion environment recognition. Beyond the control of lower-limb exoskeletons and prostheses, applications of ExoNet extend to humanoid and autonomous legged robotics.Reference: Laschowski B, McNally W, Wong A, and McPhee J. (2020). ExoNet Database: Wearable Camera Images of Human Locomotion Environments. In Preparation.摘要：近年来，计算机视觉与人工智能领域的最新进展助力研究者开发出面向下肢外骨骼与假肢的环境识别系统。然而，规模受限且私有化的训练数据集，阻碍了人类行走环境识别用图像分类算法的大规模开发与传播推广。为解决上述不足，我们构建了"ExoNet"——全球首个开源、大规模层级化的人类行走环境可穿戴相机高分辨率图像数据库。该数据库在规模与多样性上均无可比拟，共收录超过560万张涵盖不同室内外真实行走场景的图像，采集工作依托轻量化可穿戴智能手机相机系统完成，覆盖夏、秋、冬三个季节。ExoNet中约94万张图像采用12类层级标注架构完成人工标注。本数据库可通过IEEE DataPort数据仓库公开获取，为训练、开发及对比面向人类行走环境识别的下一代图像分类算法提供了前所未有的社区化研发平台。除下肢外骨骼与假肢的控制应用外，ExoNet的应用场景还可拓展至人形机器人与自主足式机器人领域。 参考文献：Laschowski B, McNally W, Wong A, and McPhee J. (2020). ExoNet数据库：人类行走环境可穿戴相机图像 [在编].