Checkpoint image vehicle retrieval method and system

Abstract

The invention discloses a checkpoint image vehicle retrieval method and system, the method constructs a checkpoint image vehicle retrieval model, and the checkpoint image vehicle retrieval model is composed of a detection network for obtaining a target vehicle image block, a vehicle key point positioning network and a vehicle image block coding network; the method comprises the following steps oftraining the checkpoint image vehicle retrieval model by using the training sample. collecting images of the checkpoint vehicles, inputting the images into the trained checkpoint image vehicle retrieval model, and retrieving from the database to obtain different images belonging to the same vehicle; According to the method, global information including camera postures, vehicle types and the like is used for assisting positioning of vehicle key points, so that accurate vehicle image blocks are obtained; The vehicle image block coding network adopts a quaternary loss function perceived by a sample space structure, negative sample information is fully explored, and the problem of limited performance improvement of the ternary loss function is solved. According to the method, the vehicle picture retrieval accuracy is effectively improved.

Description

technical field [0001] The invention belongs to the field of image vehicle retrieval, and relates to a bayonet image vehicle retrieval method and system. Background technique [0002] With the deployment of a large number of bayonet cameras, the wide application of bayonet image recognition such as vehicle flow rate monitoring, illegal driving evidence collection, and vehicle direction monitoring, vehicle retrieval from bayonet images has become a hot spot in the transportation industry. [0003] In recent years, with the wide application of deep learning, many classification and regression tasks have adopted the method of convolutional neural network on a large scale, and the method of using CNN has also achieved many successful researches on content-based image retrieval. . In image vehicle retrieval, many vehicle images only account for a part of the total image, and if the images include too many irrelevant background factors, the retrieval results will be affected. In...

AI Technical Summary

Problems solved by technology

However, when it comes to instance retrieval, CNN-based methods will encounter two problems: the first point is how to accurately locate the vehicle image block in the image; the second point is when the number of negative samples is much larger than the positive samples How to effectively use the information in the training data when the number of

Images