An image management method and device based on a multi-task machine learning model

Abstract

The application discloses an image management method and device based on a multi-task machine learning model, and relates to artificial intelligence machine learning technology. By acquiring image data; then input the target image into the shared feature expression network in the multi-task machine learning model to obtain the task output features; and input the task output features into multiple sub-task networks to obtain the recognition results; and then based on the recognition results to Image data is processed. In this way, the image management process based on machine learning is realized. Since the features used for multiple task execution are extracted through the shared feature expression network, and different subtask networks are used to perform corresponding tasks, the multi-task machine learning model can be used to achieve multiple tasks. The execution process of an image management task improves the efficiency of image management.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to an image management method and device based on a multi-task machine learning model. Background technique [0002] With the popularization of smart phones and the increasing number of mobile phone photos, more and more people have the demand task of organizing mobile phone photos in real time. These demand tasks mainly have the following salient features. First of all, the task correlation is strong; the scene category of the photo has a great relationship with the main object in the photo, and various quality indicators of the photo, such as blur and exposure, are also related to the scene where the photo was taken. In addition, photo scene recognition, object detection, and photo quality assessment all rely on photo texture and contour light features. In addition, task processing requires high model size and computational complexity. Because the model needs to be d...

AI Technical Summary

Problems solved by technology

[0004] However, due to the large amount of tasks involved in the image sorting process, designing a separate machine learning model for each task takes up a lot of system resources, and the preprocessing time of multiple models in the design process is long and the process is complicated, which affects the image quality. management efficiency

Images