Novel Low Memory Footprint DNN Models for Edge Classification of Surgeons’ Postures

Skill assessment is fundamental to enhance current laparoscopic surgical training and reduce the incidence of musculoskeletal injuries from performing these procedures. Recently, deep neural networks (DNNs) have been used improve human posture surgeons’ skills training. While they work well in lab, normally require significant computational power which makes it impossible use them on edge devices. This letter presents two low memory footprint DNN models for classifying skill levels at edge. Trained were deployed three Arm Cortex-M processors using X-Cube-AI Tensorflow Lite Micro (TFLM) libraries. Results show that CUBE-AI-based give best relative performance, footprint, accuracy tradeoffs when executed Cortex-M7.