In order to verify the performance of the YOLOv5 algorithm incorporating the SE attention module in this experiment appropriate evaluation metrics are taken to evaluate the network model and the metri

In this experiment, appropriate evaluation metrics were selected to verify the performance of the YOLOv5 algorithm incorporating the SE attention module. The precision (P), recall (R), and mean average precision (mAP) metrics were chosen to evaluate the model.

The calculation of precision P is shown in Equation (3), where TP (True Positive) represents the true cases, i.e., wearing reflective clothing in this model, and FP (False Positive) represents the false positive cases, i.e., not wearing reflective clothing.

The recall R is calculated as shown in Equation (4), where TP is the same as in Equation (3), and FN (False Negative) indicates the false negative cases, i.e., when someone is actually wearing the reflective jacket, but the model detects not wearing it.