Target recognition method, system and device in commodity image, and storage medium

A technology of target recognition and commodity image, applied in the field of image recognition, it can solve the problem of not being able to adapt to complex scenes, and achieve the effect of improving the recognition accuracy and avoiding missed detection.
CN108734162AActive Publication Date: 2018-11-02SHANGHAI CLOBOTICS TECHNOLOGY CO LTD

Patent Information

Authority / Receiving Office
CN · China
Current Assignee / Owner
SHANGHAI CLOBOTICS TECHNOLOGY CO LTD
Publication Date
2018-11-02

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The present invention provides a target recognition method, system and device in a commodity image, and a storage medium. The method comprises: S1: acquiring a plurality of images, inputting the images into a pre-trained scenario classification model, and determining an application scenario corresponding to an image; S2: selecting a pre-trained ROI area detection model according to the applicationscenario corresponding to the image, and detecting ROI areas on the image; S3: correcting the image with the detected ROI areas to generate a corrected image; and S4: inputting the corrected image into a pre-trained target recognition model to perform recognition of each commodity area on the image. According to the technical scheme of the present invention, before the recognition of the commodity area, scenario classification, quality adjustment, ROI area recognition, and image correction are sequentially performed in an innovative manner, and finally recognition of the commodity area is performed, so that the recognition accuracy of the commodity area is remarkably improved, the recognition accuracy of the commodity is increased from 60% to 95%, and the application of the commodity recognition technology in new retail areas is promoted.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to image recognition technology, in particular to a method, system, device and storage medium for object recognition in commodity images. Background technique

[0002] With the increasing computing power of the hardware part of computers and smartphones, machine learning and computer vision technology has achieved rapid development in recent years. Many researchers have done a lot of work to develop machine learning and computer that can assist people in daily tasks. vision algorithm.

[0003] In the field of new retail, it is necessary to regularly collect shelf images through mobile phones, and collect the collected images into the background system. The acquired system uses image processing technology to obtain the type, quantity and location information of each commodity in the image. After the generated data is aggregated into a report and the image data is updated regularly. It is possible to fully control the con...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More