Face detection method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a face detection method and device, electronic equipment and a storage medium, and belongs to the technical field of machine learning. The method comprises the following steps: obtaining a normalized face image; feature extraction is carried out on the normalized face image by using a convolutional neural network to obtain a plurality of face detection results, a face center point modeling scheme is adopted in face detection, and cavity convolution and multi-scale multi-stage supervised training are used during feature extraction; and performing non-maximum suppression on the plurality of face detection results to obtain a final face detection result. The face detection speed is high, and the detection effect is good.

Description

technical field [0001] The present invention relates to the technical field of machine learning, in particular to a face detection method, device, electronic equipment and storage medium. Background technique [0002] CNN (Convolutional Neural Networks, convolutional neural network) has made great progress in the field of pattern recognition, especially in the field of image face recognition, the recognition effect can far exceed the discrimination effect of human eyes. Since CNN can adaptively obtain image features and obtain better classification results, CNN-based face detection has also made a major breakthrough. However, many face detection methods model the face as a rectangular frame, which is less robust to the pose of the face. Moreover, many existing network structures increase the receptive field of feature maps by downsampling, but downsampling will reduce the resolution of features, which is not conducive to better detection of faces. [0003] According to dif...

AI Technical Summary

Problems solved by technology

However, its speed depends on the number of candidate windows and the size of the classification network. The larger the number of candidate windows and the deeper the classification network, the slower the detection speed.

[0008] In the existing CNN-based face detection technology, the one-stage face detection method does not need to obtain a proposal, so it needs a deeper network to obtain a better detection effect, and the face search space is wider, resulting in difficulties in training. The requirements are relatively high; the detection speed of the two-stage face detection method is limited by the number of proposals. In order to obtain better detection results, the number of candidate windows is often large, and there will be overlapping parts between candidate windows resulting in a large number of repetitions. calculation, thereby slowing down the speed of detection

Images