fig1
Figure 1. Processing pipeline of the automated assessment system. Recorded-video or live-camera input is sampled at the selected frame rate, letterbox-padded to 640 × 640 pixels, and passed to the YOLOv8n detector; confidence filtering and non-maximum suppression yield object and task-state detections, which a deterministic rule-based state machine combines with retained task history to update task state, detect errors, and render a pass/fail decision with on-screen learner feedback. RGB: Red, green, blue; px: pixels; YOLO: You Only Look Once; ONNX: Open Neural Network Exchange; IoU: intersection of union score; FLS: Fundamentals of Laparoscopic Surgery.





