Target re-identification method and device, computer equipment and storage medium

Abstract

The invention relates to a target re-identification method and device based on deep learning, computer equipment and a storage medium. The target re-identification method comprises the following steps: identifying a target object picture and a view angle of each to-be-detected picture in a to-be-detected picture set, and determining a same view angle picture with the same view angle as the targetobject picture in the to-be-detected picture set and a different view angle picture with a different view angle from the target object picture; determining a first feature distance between the targetobject picture and the same view angle picture, and a second feature distance between the target object picture and the different view angle pictures; according to the first feature distance, determining pictures including the target object under the same view angle; and according to the second feature distance, determining pictures including the target object under different view angles. By adopting the target re-identification method, the pictures comprising the same target object under the same visual angle and different visual angles can be determined respectively, and the problem that thesame target object cannot be accurately determined due to large shooting visual angle difference is effectively solved, and the accuracy of target re-identification work is improved.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a target re-identification method, device, computer equipment and storage medium. Background technique [0002] With the development of computer technology, target re-identification technology has appeared. Object re-identification is used to determine, for a given object, whether there is an object with the same identity at another time or perspective. Among them, the target can be a vehicle or a pedestrian, and target re-identification is used to match vehicles or pedestrians with different viewing angles or time periods in video surveillance. [0003] In traditional technology, it is usually determined whether the same target exists by using the appearance information of the target. However, since the appearance of the target is easily affected by external conditions and factors of the target itself, the appearance information of the same target is inconsistent in...

AI Technical Summary

Problems solved by technology

[0003] In traditional technology, it is usually determined whether there is the same target by using the appearance information of the target. However, because the appearance of the target is easily affected by external conditions and factors of the target itself, the appearance information of the same target is inconsistent in different environments.

For example, if the same pedestrian object is under different lighting conditions or different postures, there are large differences in the captured images of the same pedestrian, and it is impossible to determine them as the same target

For vehicles, under different viewing angles or different resolutions of video images, the appearance information of the same vehicle is different, and it is also impossible to effectively realize vehicle re-identification

Therefore, in the traditional re-identification method, the appearance information of the target is easily affected. In various practical application scenarios, the accuracy of target re-identification is low.

Images