Neural network general pre-processing acceleration method and device, equipment and medium

By generating general operating streams and automatic data slices, the problem of insufficient preprocessing performance of intelligent digital signal processors is solved, and efficient preprocessing acceleration is achieved.

CN119722427APending Publication Date: 2025-03-28HUNAN GREAT WALL GALAXY TECH CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202411811324.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-10
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

When using intelligent digital signal processors for preprocessing in the prior art, the versatility and acceleration performance of preprocessing need to be improved.

Method used

By obtaining and parsing configuration files, a general operation flow is generated, the preprocessing of different neural networks is divided into multiple sub-operations, and the data slice is automatically completed based on the number of cores of the intelligent digital signal processor, the preprocessing calculation is completed through a single IO read and write operation, and the storage index table is automatically generated.

Benefits of technology

The diversity and efficient acceleration of preprocessing are achieved, greatly reducing the time for data reading, writing and processing, and improving the preprocessing performance of using smart digital signal processors.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to a neural network general pre-processing acceleration method, device, equipment and medium, and the method comprises the steps: reasonably dividing various pre-processing corresponding to different neural networks into a plurality of sub-operations, and determining picture input formats and network data input formats for different neural networks; combination of a plurality of sub-operations is completed in a configuration file mode, a universal operation stream is generated, and picture data is processed according to the generated operation stream, so that diversity of pre-processing is realized, redundant calculation is avoided, and the intelligent digital signal processor end is provided with a plurality of sub-operations. Multi-core parallel processing is realized by automatically fragmenting the image data and reading and writing all the operation streams at one time, and the storage index table is automatically generated to realize storage of the processed data, so that the data reading, writing and processing time is greatly shortened; and the universality and acceleration performance of pretreatment when the intelligent digital signal processor is used for calculation are effectively improved.
Need to check novelty before this filing date? Find Prior Art