Safety helmet detection method and device based on YOLOv7-WFD model

The invention discloses a safety helmet detection method and device based on a YOLOv7-WFD model, and the method comprises the steps: obtaining an original image of a construction site in advance, and carrying out the preprocessing of the image; constructing a YOLOv7-WFD model: in the original YOLOv7 model, replacing the last ELAN module in the trunk with an ELAN-D module, that is, replacing a part of CBS in the original ELAN module with DBS; the DBS module is composed of a deformable convolutional layer DCN, a normalization layer and a SiLU activation function; in addition, an UpSample module in the head is replaced by a CARAFE up-sampling operator, so that the visual field range of the model is expanded; a YOLOv7-WFD model is trained, and the performance of the YOLOv7-WFD model is evaluated by using a Wise-IoU loss function combined with a dynamic non-monotonic focusing mechanism as bounding box regression loss; and inputting an image to be detected into the trained YOLOv7-WFD model, and outputting a detecti