Picture identification method and device

Abstract

The embodiment of the invention provides a picture recognition method and device. The method comprises the following steps: carrying out object identification on a first picture and a second picture to respectively obtain a plurality of first detection frames and a plurality of second detection frames; for any first detection frame, determining the first detection frame and the second detection frame with the position matching relationship as a first abnormal frame and a second abnormal frame respectively; determining a position overlapping area of the first abnormal frame and the second abnormal frame; determining a first similarity between a first sub-image corresponding to the position overlapping area in the first picture and a second sub-image corresponding to the position overlapping area in the second picture; and if the first similarity meets a first set condition, updating the second identification result of the second abnormal frame to the first identification result of the first abnormal frame. Thus, it can be ensured that the recognition results of the first picture and the second picture are consistent for the same commodity, the payment amount of the user can be prevented from being calculated wrongly, and better experience is brought to the user.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of computer vision, and in particular, to a picture recognition method, device, computing device, and computer-readable storage medium. Background technique [0002] In recent years, with the rapid development of deep learning technology, many smart vending machines that are different from traditional vending machines have appeared on the market. This kind of smart vending machine is the application of deep learning technology in computer vision. Figure 1A shows an intelligent vending machine, which is divided into 4-5 layers of shelves, and the height of each layer can be equal or different, and is used to place retail goods. There is a camera on the top of each floor to take pictures of the goods on this floor. In order to make the camera capture the goods in all positions of the floor as completely as possible, it is necessary to install the camera on the top of each shelf and center...

AI Technical Summary

Problems solved by technology

[0003] However, the target detection algorithm has a certain error rate. If there is a product, the type detected in the two target detections has changed, which will lead to a wrong calculation of the deduction amount. For example, when the user opens the cabinet door In the process of closing the cabinet door, no product was taken away, but when the smart vending machine detects the target of the pictures before and after the user opens and closes the cabinet door, the same product is recognized as mineral water before the user opens the door, and the user closes the door. If it is recognized as fruit juice behind the cabinet door, after calculation, it is determined that a bottle of mineral water is missing, and the deduction amount of a bottle of mineral water will be displayed for the user

The occurrence of such target detection errors leads to lower accuracy of automatic settlement results, which greatly reduces the user's shopping experience, and also affects the inventory management of commodities.

Images