Novel deep learning vehicle detection method and system

By introducing attention mechanism and adaptive feature fusion model into the yolov5s model, the shortcomings of vehicle detection algorithm in real time and accuracy are solved, and more efficient vehicle detection effects are achieved.

CN119068436APending Publication Date: 2024-12-03CHANGCHUN UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411179280.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-12-03

Smart Images

  • Figure CN119068436A_ABST
    Figure CN119068436A_ABST
Patent Text Reader

Abstract

The invention provides a novel deep learning vehicle detection method and system, and the method comprises the steps: obtaining an original image set of a target vehicle, and carrying out the preprocessing of the original image set; constructing an improved yolov5s model, and introducing an attention mechanism model and an adaptive feature fusion model into the model; and inputting the processed original image set into the improved yolov5s model, and carrying out image feature extraction by adopting a single-stage vehicle detection technology. Compared with the prior art, according to the technical scheme of the invention, the attention model and the improved feature fusion model are added in a traditional model, so that the accuracy and the real-time performance of vehicle detection are effectively improved, and the detection capability of an algorithm model on targets with different scales is effectively improved.
Need to check novelty before this filing date? Find Prior Art