Non-motor vehicle identification method, system and device

Abstract

The invention discloses a non-motor vehicle identification method, system and device, and the method comprises the steps of obtaining a target image; generating an object segmentation result of the target image, wherein the object segmentation result is used for representing a person and / or a non-motor vehicle to be detected in the target image; extracting image features of the target image, and identifying attention features of the image features in an area limited by the object segmentation result; and determining a target identifier of a non-motor vehicle in the target image according to the attention feature, so as to track the non-motor vehicle based on the determined target identifier. According to the technical scheme provided by the invention, the re-identification precision of the non-motor vehicle can be improved.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a recognition method, system and device for non-motor vehicles. Background technique [0002] In the intelligent monitoring scenario, the demand for re-identification of non-motor vehicles is increasing. When the non-motor vehicle is re-identified, the same non-motor vehicle target captured by another camera can be searched for the non-motor vehicle target captured by one camera, so as to realize the tracking of the non-motor vehicle target. [0003] In practical applications, non-motor vehicles in surveillance images are usually combined with human bodies, for example, people will sit on non-motor vehicles, or people will walk with the support of non-motor vehicles, which will make it difficult to track non-motor vehicles larger. According to the existing re-identification technology, it is difficult to distinguish a human body from a non-motorized vehicle, resulting...

AI Technical Summary

Problems solved by technology

[0003] In practical applications, non-motorized vehicles in surveillance images are usually combined with human bodies, for example, people will sit on non-motorized vehicles, or people will walk with the help of non-motorized vehicles, which will make it difficult to track non-motorized vehicles larger

According to the existing re-identification technology, it is difficult to distinguish the human body from the non-motor vehicle, resulting in low accuracy of re-identification

Images