Target positioning model training method, target positioning method and device

Abstract

The invention provides a target positioning model training method, target positioning method and device, which can be applied to the technical field of artificial intelligence. The target positioning model training method includes: acquiring a sample data set; inputting each image sample into a pixel feature extraction layer of an initial model, outputting the pixel feature of each pixel in the image data; inputting the first feature data and the second feature data The activation map generation layer of the initial model outputs the foreground activation map and the background activation map; the third feature data and the fourth feature data are respectively input into the perception feature extraction layer of the initial model, and the foreground perception feature and background perception feature are output; the fifth feature The data and the sixth feature data are respectively input into the classification layer of the initial model, and the classification result is output; and the model parameters of the initial model are adjusted according to the classification result and the image category label, and the trained target positioning model is obtained.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a target positioning model training method, target positioning method, device, equipment and medium. Background technique [0002] Traditional target localization methods are usually based on fully supervised learning, which uses manually annotated bounding boxes to train models for target localization. However, such annotated data is expensive and time-consuming, which limits the practicability of target localization. [0003] With the development of artificial intelligence technology, weakly supervised learning algorithms have gradually attracted attention. However, in the process of implementing the present invention, the inventor found that when the existing weakly supervised learning algorithms are applied to target positioning, there is a problem of incomplete positioning. SUMMARY OF THE INVENTION [0004] In view of the above problems, the present invent...

AI Technical Summary

Problems solved by technology

However, such labeled data is expensive and time-consuming, limiting the usefulness of target localization

[0003] With the development of artificial intelligence technology, weakly supervised learning algorithms have gradually attracted attention. However, the inventor found in the process of implementing the present invention that: when the existing weakly supervised learning algorithms are applied to target positioning, there is a problem of incomplete positioning.

Images