A construction method for a multi-scale lightweight face detection model and a face detection method based on the model, the method comprising: A: constructing a lightweight feature pyramid network module on the basis of separation anti-aliasing convolution and channel pooling technology, and introducing the feature pyramid network module into a lightweight face detection convolutional neural network model to form a multi-scale lightweight face detection model; and B: acquiring a specified number of face digital images marked with face positions and sizes so serve as a training data set, and iteratively training the multi-scale lightweight face detection model by using the training data set so as to obtain a trained multi-scale lightweight face detection model. Thus, the scale of the detection model may be effectively reduced while the accuracy of face detection is increased, thereby adapting to a face detection task of an embedded platform having limited resources.
展开▼