Deep learning model conversion deployment method and device, storage medium and electronic equipment

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a deep learning model conversion deployment method, a deep learning model conversion deployment device, a storage medium and electronic equipment. The deep learning model conversion deployment method comprises the steps of obtaining a to-be-processed model file obtained by performing model training by using a first deep learning framework; calling a deconvolution function to perform up-sampling operation on the to-be-processed model file to obtain an initial model file; performing data normalization processing on the initial model file by adopting a normalization function to obtain an intermediate model file; and performing linear correction on the intermediate model file by using an activation function to obtain a target model file which is suitable for processing in a second deep learning framework. According to the deep learning model conversion deployment method provided by the invention, the technical problems of large development amount, poor precision and high cost when a pytorch model is converted into a TensorRT model for deployment in the prior art can be solved.

Description

technical field [0001] The present disclosure relates to the technical field of artificial intelligence, and in particular to a method for converting and deploying a deep learning model, a device for converting and deploying a deep learning model, a storage medium, and electronic equipment. Background technique [0002] In engineering applications, the pytorch model can be converted to the TensorRT engine to achieve the effect of computing acceleration. However, when the pytorch model is converted to the 5.0.2.6 version of the TensorRT engine file, the upsampling operation torch.nn.Upsample() cannot be successfully converted to the TensorRT engine. [0003] Among the existing technologies, one is to solve it by developing a customized plug-in of TensorRT5.0.2.6, but this method has a large amount of engineering development and the steps are relatively cumbersome; or first generate the intermediate transition model onnx from pth, and then generate from onnx TensorRT5.0.2.6 m...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present disclosure is to provide a deep learning model conversion and deployment method, a deep learning model conversion and deployment device, storage media and electronic equipment, aiming to solve the problems of large amount of development and problems existing in the prior art when converting a pytorch model to a TensorRT model. Technical problems of poor precision and high cost

Images