Image processing model quantification method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an image processing model quantification method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining image processing models of a plurality of storage points, and obtaining a plurality of to-be-quantized models; performing model quantization on each to-be-quantized model to obtain each quantized model; obtaining a quantitative test picture, and analyzing the quantitative test picture by using each quantitative model to obtain a test result of each quantitative model; and determining a target quantification model based on the test result of each quantification model. Model quantization is carried out on the image processing model of multiple storage points in one training, the problem that the accuracy of a quantization model obtained by selecting a single model for quantization due to random distribution of model parameters is low can be reduced, and the success rate of model quantization can be increased; and manual parameter adjustment on the image processing model is not needed, so that the manual workload is reduced, and a good foundation is laid for batch model quantification.

Description

technical field [0001] The present application relates to the technical field of deep learning, and in particular to an image processing model quantification method, device, electronic equipment and storage medium. Background technique [0002] With the development of artificial intelligence technology, deep learning models are increasingly used in image processing scenarios. Deep learning model quantization refers to the process of approximating the tensor data of continuous floating-point model weights to a limited number of discrete values ​​with a lower loss of inference accuracy. It uses a data type with fewer digits for approximation The process of representing 32-bit limited-range floating-point data, while the input and output of the deep learning model is still floating-point data, so as to reduce the size of the deep learning model, reduce the hardware consumption of the deep learning model, and speed up the inference speed of the deep learning model, etc. Purpose...

AI Technical Summary

Problems solved by technology

However, using the above method, the accuracy of the obtained quantitative model is random, and it cannot be guaranteed to obtain a quantitative model with high accuracy.

Images