Computer Vision for Clinical Gait Analysis: A Gait Abnormality Video Dataset

TL;DR

The paper tackles the scarcity of clinically annotated RGB video data for clinical gait analysis by introducing GAVD, the Gait Abnormality in Video Dataset, a large, publicly accessible collection of normal, abnormal, and pathological gait sequences with expert annotations. It demonstrates the utility of GAVD by fine-tuning region-based action recognition models (TSN and SlowFast), achieving high accuracy on GAVD and revealing challenges in cross-dataset generalization and camera-view effects. The work also provides a CASD validation dataset and GPJATK and CASD test sets to benchmark model robustness, while offering insights into the need for view-invariant, frame-wise gait analysis and future directions such as segmentation-based attention and temporal-normalization techniques. Overall, GAVD fills a critical gap in CGA research by enabling clinically meaningful, video-based gait analysis and supporting the development of more generalizable, interpretable gait-abnormality detection methods.

Abstract

Clinical gait analysis (CGA) using computer vision is an emerging field in artificial intelligence that faces barriers of accessible, real-world data, and clear task objectives. This paper lays the foundation for current developments in CGA as well as vision-based methods and datasets suitable for gait analysis. We introduce The Gait Abnormality in Video Dataset (GAVD) in response to our review of over 150 current gait-related computer vision datasets, which highlighted the need for a large and accessible gait dataset clinically annotated for CGA. GAVD stands out as the largest video gait dataset, comprising 1874 sequences of normal, abnormal and pathological gaits. Additionally, GAVD includes clinically annotated RGB data sourced from publicly available content on online platforms. It also encompasses over 400 subjects who have undergone clinical grade visual screening to represent a diverse range of abnormal gait patterns, captured in various settings, including hospital clinics and urban uncontrolled outdoor environments. We demonstrate the validity of the dataset and utility of action recognition models for CGA using pretrained models Temporal Segment Networks(TSN) and SlowFast network to achieve video abnormality detection of 94% and 92% respectively when tested on GAVD dataset. A GitHub repository https://github.com/Rahmyyy/GAVD consisting of convenient URL links, and clinically relevant annotation for CGA is provided for over 450 online videos, featuring diverse subjects performing a range of normal, pathological, and abnormal gait patterns.

Figures (4)

• Figure 1: Example frames representing a variety of subjects from GAVD dataset
• Figure 2: Distribution of Gait Datasets by Computer Vision Task
• Figure 3: Distribution of Frames per Subclass in GAVD
• Figure 4: Person-centric Camera View