abstract |
The invention discloses a multi-scale fast face detection method based on convolutional neural network feature fusion, comprising the following steps: Step 1: Based on the model structure of the SSD fast target detection method, the feature extraction method in the SSD is transformed, and feature fusion is added Method, obtain the improved detection model; step 2: carry out training for face detection to the improved detection model in step 1, and obtain the trained deep neural network model; step 3: use the trained deep neural network in step 2 The model calculates the image to be detected and obtains the output result of the model. The invention can quickly identify the human face in the image and precisely locate the position of the human face, thereby separating the human face from the complex background. Provides the basis for authentication and tracking of people in images. |