Data processing method and device, data processing equipment and storage medium

Abstract

A data processing method comprises the steps: reading scheduling information from a first system memory through task scheduling hardware, wherein the scheduling information comprises public configuration information and a task descriptor, and the public configuration information is matched with the task descriptor; determining a target computing module needing to work in the plurality of hardwarecomputing modules through the task scheduling hardware according to the public configuration information; allocating parameters in the task descriptor to the target computing module through the task scheduling hardware; starting the target calculation module; and executing the calculation operation of the target calculation module according to the parameters. The invention further provides a dataprocessing device, data processing equipment and a storage medium. According to the invention, the universality and flexibility of the customized hardware computing module can be improved.

Description

technical field [0001] The present invention relates to the field of image technology, in particular to a data processing method, device, data processing equipment and storage medium. Background technique [0002] In many application scenarios, hardware computing modules are required to perform calculation operations of related algorithms, so as to realize information extraction. Train an image object tracker, use the image object tracker to detect whether the predicted position of the next frame is an object, and then use the new detection result to update the training set to update the image object tracker. This requires customizing multiple hardware computing modules specifically for image object trackers. [0003] However, for other algorithms, such as face recognition algorithms, due to different algorithms, only some of the above-mentioned multiple hardware computing modules may be used, and the hardware computing modules customized for the above-mentioned KCF algorit...

AI Technical Summary

Problems solved by technology

[0003] However, for other algorithms, such as face recognition algorithms, due to different algorithms, only some of the above-mentioned multiple hardware computing modules may be used, and the hardware computing modules customized for the above-mentioned KCF algorithm cannot be used in face recognition algorithms. It is reused, and the face recognition algorithm needs to be re-customized. The above-mentioned customized multiple hardware computing modules have poor versatility

Images