Commodity positioning identification method, device and equipment and storage medium

Abstract

The invention discloses a commodity positioning identification method. The method comprises the steps of obtaining an input to-be-identified image; Inputting the to-be-identified image into a pre-trained neural network model, and extracting an area corresponding to a commodity in the to-be-identified image; Neural network model trained in advance is Faster- An RCNN network model; Intercepting an area corresponding to the commodity as a subgraph, and inputting the subgraph into a pre-trained image classification neural network model; Pre-trained image classification neural network model as VGG-A Net network model; And calculating the probability that the sub-image belongs to each commodity category through an image classification neural network model, and identifying the category of the commodity in the to-be-identified image. According to the method, video memory occupation can be reduced, the detection speed can be increased by about ten times, and meanwhile the recognition accuracycan be improved. In addition, the invention further provides a commodity positioning identification device and equipment with the technical effects and a computer readable storage medium.

Description

technical field [0001] The present invention relates to the technical field of computer vision, in particular to a commodity location recognition method, device, equipment and computer-readable storage medium. Background technique [0002] The use of deep learning and computer vision technology to identify goods is an important part of unmanned retail and the core technology for realizing intelligent settlement desks. The current mainstream deep learning-based target detection algorithms include Faster-RCNN, SSD, YOLO, etc. The two-step algorithm Faster-RCNN has precise positioning and recognition effects. However, when the number of categories is large, the algorithm will occupy a large amount of video memory, the recognition speed will decrease proportionally, and the recognition accuracy is difficult to control. The one-step algorithm SSD and YOLO have significantly improved the recognition speed, but the positioning is not accurate enough, and the recognition accuracy i...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a commodity location recognition method, device, equipment and computer-readable storage medium to solve the problems in the

Images