Model construction method and device, terminal device and medium

Abstract

The method is suitable for the technical field of model construction, and provides a model construction method, a model construction device, a terminal device and a medium. The model construction method comprises the following steps: determining a convolution layer queue from a trained face detection model; then, based on the channel weight of each feature channel of each convolution layer in theconvolution layer queue and a preset channel deletion proportion, deleting the feature channels in the convolution layer queue; arranging the convolution layers according to the arrangement sequence of the convolution layers in the convolution layer queue; performing channel deletion on each convolution layer in sequence. According to the method, the convolution layer is not deleted in the new face detection model, and only part of channels in the convolution layer are deleted, so that the resource occupancy rate when the new face detection model is operated can be reduced under the conditionof ensuring that the face detection accuracy is not changed, and meanwhile, the face image recognition efficiency is improved.

Description

technical field [0001] The present application belongs to the technical field of model building, and in particular relates to a method, device, terminal equipment and computer-readable storage medium for building a model. Background technique [0002] Face detection technology is a biometric technology for identification based on human facial feature information. At the same time, face detection technology is also one of the most widely used technologies among various biometric technologies. For example, the unlocking of terminals, payment verification in consumer activities, or access control attendance, etc., are all applied to face detection technology. [0003] When recognizing face images, most of them use face image recognition models including neural networks to recognize the collected face images. Although the existing face image recognition models are equipped with a relatively complex feature extraction network layer, which can be used to extract various features...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the embodiment of the present application provides a method, device, terminal device and computer-readable storage medium for constructing a model, so as to solve the problem of low recognition efficiency of a face detection model in the existing face image recognition process. The problem

Images