Pseudocode for YOLOv5 Detect Part (Important Steps)

Loading system libraries, setting the system environment.         Input: The CWD picture of Mixed signals.         Step 1:The size of the zoom image is [640 640]         Step 2: Conf_value = 0.3 ## Boxes whose confidence is higher than 30% will be reserved.         Step 3: Iou_value = 0.4 ## IOU boxes whose value is higher than this value can be reserved.         Step 4: max_det = 10; ## Maximum number of targets. This value can change depending on the testing environment.

Predicting part         Step 1:Use blank picture (0 matrix) prediction to accelerate the prediction process.Import single-label and multi-label datasets.         Step 2: im/= 255 # 0–255 to 0.0–1.0 # Normalize the image.                 if len(im.shape) == 3:                 im = im[None] ## Add the 0th dimension.         Step 3: pred = model(im, augment = augment, visualize = visualize)                 pred = non_max_suppression(pred,conf_thres,iou_thres,classes, agnostic_nms, max_det = max_det) ##Prediction box save and NMS         Step 4: if len(det):                 # Rescale boxes from img_size to im0 size.                 det [:, :4] = scale_coords(im.shape [2:], det [:, :4], im0.shape).round()                 # Resize the annotated bounding_box to the same size as the original image (because the original image has been scaled up and down during training)                 # Print results     Output the predicted time for each image.                 The accuracy distribution is returned according to the result of each prediction, and the lowest prediction accuracy category is output.

Output: Mixed signal CWD image with predict box.                 The target category.                 Degree of confidence.                 The box position.