Skip to main content
Fig. 3 | Journal of NeuroEngineering and Rehabilitation

Fig. 3

From: Automated freezing of gait assessment with marker-based motion capture and multi-stage spatial-temporal graph convolutional neural networks

Fig. 3

Toy example to visualize the IoU computation and segment classification. The predicted FOG segmentation is visualized in pink, the experts’ FOG segmentation in gray, and the color gradient visualizes the overlap between the predicted and experts’ segmentation. The intersection is visualized in orange and the union in green. If a FOG segment’s IoU (intersection divided by union) crosses a predetermined threshold it is classified as a TP, if not, as a FP. For example, the FOG segment with an IoU of 0.42 would be classified as a FP. Given that the number of correctly detected segments (n = 0) is less than the number of segments that the experts demarcated (n = 1), there would be 1 FN

Back to article page