The work in this paper focuses on the role of machine learning in assessing the correctness of a human motion or action. This task proves to be more challenging than the gesture and action recognition ones. We will demonstrate, through a set of experiments on a recent dataset, that machine learning algorithms can produce good results for certain actions, but can also fall into the trap of classifying an incorrect execution of an action as a correct execution of another action.
Rehabilitation Exercise Quality Assessment and Feedback Generation Using Large Language Models with Prompt Engineering
KITE Research Institute（KITE研究 institute）；Toronto Rehabilitation Institute（多伦多康复研究所）；University Health Network（大学健康网络）；Faculty of Applied Science and Engineering（应用科学与工程学院）；University of Toronto（多伦多大学）；College of Engineering and Technology（工程与技术学院）；American University of the Middle East（中东美国大学）
GAITEX: Human motion dataset of impaired gait and rehabilitation exercises using inertial and optical sensors
AI for Sensor Data Analytics Research Group（人工智能传感器数据解析研究组）；Ulm University of Applied Sciences（乌尔姆应用科学大学）；Biomechatronic Research Group（生物机械研究组）；Institute of Computer Science（计算机科学研究所）
Towards nation-wide analytical healthcare infrastructures: A privacy-preserving augmented knee rehabilitation case study
AUT University（奥克兰理工大学）；Institute of Biomedical Technologies (IBTec)（生物医学技术研究所（IBTec））；Sports Performance Research Institute New Zealand (SPRINZ)（新西兰运动表现研究所（SPRINZ））；Ovidius University of Constanta（康斯坦察奥维迪乌斯大学）
本文提出一种隐私保护的膝关节康复分析方案，利用 Google MediaPipe 姿态估计将视频转为时间序列数据，并通过自适应算法以91.67%-100%准确率识别运动，为全国性可解释医疗平台提供开源基础。
HRTR: A Single-stage Transformer for Fine-grained Sub-second Action Segmentation in Stroke Rehabilitation
Department of Electrical and Computer Engineering, University of Kentucky（电气与计算机工程系，肯塔基大学）；College of Medicine, University of Kentucky（医学学院，肯塔基大学）；Department of Electrical and Computer Engineering, University of California, Davis（电气与计算机工程系，加州大学戴维斯分校）
提出单阶段 High Resolution Temporal Transformer（HRTR），用于中风康复中细粒度亚秒级动作的时间定位与分类，无需多阶段流程或后处理，并在多个数据集上取得最优 Edit Score。
