Vision-based biomechanics has enabled valuable tools for movement analysis, supporting applications such as rehabilitation monitoring, gait assessment, and neurological evaluation. However, these advances have largely been developed and validated in well-resourced environments, limiting their accessibility in resource-limited settings (RLS) where infrastructure, connectivity, and technical capacity remain constrained. In this paper, we examine why many current vision-based biomechanics pipelines fail to translate effectively to such contexts, identifying key barriers and important ethical considerations. We then propose a selection of design strategies aimed at improving applications for RLS, advocating for greater resourcefulness and practicality. We aim to inspire future directions and highlight how vision-based biomechanics can be redesigned to operate under constrained conditions. Finally, we propose that progress in the field should be evaluated not only by technical performance, but by the ability of systems to deliver meaningful and equitable impact in the environments where they are most needed.