Real-time target detection method applied to edge device

The invention relates to the technical field of computer vision, in particular to a real-time target detection method applied to edge equipment. Comprising the steps of constructing a target detection model; performing data enhancement processing on the data set; inputting the image after data enhancement into a target detection model; comparing by utilizing a loss function, calculating an error between a prediction result and the label data, and optimizing parameters of the target detection model by utilizing an optimizer through the error and gradient information of the target detection model; and converting the trained target detection model into a quantitative model, deploying the quantitative model on an edge device, and performing target detection on the video stream in the use scene. According to the invention, through the Rep-ELAN structure which is lighter and can be reparameterized and the decoupling detection head, the accuracy of the detector can be greatly improved while the calculation speed is