A method for identifying and detecting intelligent vending products
A detection method and product technology, which is applied in the field of intelligent vending product identification and detection, can solve the problems of visual recognition technology to increase consumer experience and save costs, and achieve the effects of increasing consumer experience, reducing volume, and good generalization ability
Active Publication Date: 2021-03-02
德明通讯(上海)股份有限公司
View PDF2 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The above method proposes a method and device for managing the inventory of autonomous vending machines through image recognition. In essence, it does not use visual recognition technology to increase consumer experience and save costs, nor does it change the above-mentioned traditional retailing methods of autonomous vending machines.
Method used
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
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
example
[0068] Example: Identify "Nutrition Express" products:
[0069] Using the Faster R-CNN in the embodiment of the present invention, combined with the method of cyclic deep learning training data sets, a total of 10 types of beverage bottle commodities were collected, including 10,000 pictures of each type with a total of 100,000 pictures, and divided into 10 groups, each group The data is approximately uniform, and one of the groups is marked with yolo_mark;
[0070] Put a set of marked data into the established Faster R-CNN for training, use the Darknet framework to set the initial learning rate of 0.1 for the first set of data for 30,000 training times, then reduce the learning rate to 0.01 for 50,000 training times, activate The function is set to leaky Relu;
[0071] After the training is completed, the saved network weight file will be obtained, which is used to automatically mark the next group of pictures;
[0072] Change the mislabeled product category and adjust the ...
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 PUM
Login to View More Abstract
The invention provides a method for identifying and detecting intelligent vending commodities, which includes: establishing commodity picture data sets for identifying items, grouping each group with the same number of categories, selecting a group for marking; training the marked group of picture data Established Faster R-CNN; use the trained Faster R-CNN weight value to automatically mark the next group; filter and correct the marking results, and add the training set for training until all image data training is completed; if you need to increase the image data set, Add the type of product recognition, and repeat the above steps until the training is complete. The invention realizes the in-depth learning of autonomous vending machine commodity cycle, the transformation from few-sample weak supervision learning to multi-sample strong supervision learning, and can freely select and identify the category of commodities according to the actual situation, reduces the complex workload of image labeling, and improves the depth Faster R‑CNN detection performance reduces the risk of overfitting.
Description
technical field [0001] The invention relates to the technical field of intelligent vending and visual target detection and recognition, in particular to a method for identifying and detecting intelligent vending commodities. Background technique [0002] In recent years, intelligence has become the focus of attention in various fields at home and abroad, and image recognition, as an important research field in intelligence, has also attracted much attention. In the retail industry, unmanned vending machines have appeared in many public places such as subways, parks and schools due to their small size, easy placement, safety and convenience, and have been widely recognized by consumers. [0003] The traditional vending machine sales methods mainly include: First, the customer selects a product on the keyboard of the vending machine and presses the corresponding button, and then chooses cash or non-cash payment. After the payment is completed, the consumer can Take the goods ...
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 Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/32G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/00G06V10/25G06V10/462G06N3/044G06N3/045G06F18/2155G06F18/2415
Inventor 后士云黄书宝亢建卫
Owner 德明通讯(上海)股份有限公司