Fast Hash Vehicle Retrieval Method Based on Multi-task Deep Learning

Abstract

A fast hash vehicle retrieval method based on multi-task deep learning, including multi-task deep convolutional neural network for deep learning and training recognition, segmented compact hash codes and examples for improving retrieval accuracy and retrieval method practicability A feature fusion method for features, a locality-sensitive hash reordering algorithm for improving retrieval performance, and a cross-modal retrieval method for improving the robustness and accuracy of retrieval engines. First, a multi-task deep convolutional network segmentation learning hash code method is proposed, which combines image semantics and image representation, uses the connection between related tasks to improve retrieval accuracy and refines image features, and at the same time minimizes image Encoding makes the learned vehicle features more robust; secondly, the feature pyramid network is selected to extract the instance features of the vehicle image; then, the extracted features are retrieved by using the local sensitive hash reordering method; finally, the unobtainable The special case of querying vehicle object images employs a cross-modal assisted vehicle retrieval approach.

Description

technical field [0001] The invention relates to the application of artificial intelligence, digital image processing, convolutional neural network and computer vision in the field of public security, and belongs to the field of intelligent transportation. Background technique [0002] Today, with the rapid development of smart cities and intelligent transportation, the demand for large-scale image monitoring and video database vehicle identification and vehicle retrieval in public security systems has increased dramatically. [0003] In the prior art, the vehicle retrieval method is mainly to extract the license plate information of the target vehicle. Then, the motor vehicle to be retrieved is retrieved according to the license plate information. The general specific method is to identify the license plate number of the vehicle from the monitoring image, and then identify whether there is a motor vehicle with the license plate number in other monitoring images. Although t...

AI Technical Summary

Problems solved by technology

[0015] Aiming at how to efficiently utilize massive video data generated in the field of public security and improve the efficiency of vehicle retrieval in the era of big data, the present invention proposes a fast hash retrieval method based on multi-task deep learning, which effectively utilizes the gap between detection and recognition tasks. Relevance and diversity of bayonet vehicle basic information to achieve the purpose of real-time retrieval; finally provide a multi-task deep learning fast hash vehicle retrieval method with high retrieval accuracy and good robustness

Images