The invention relates to a mMedical band-aid flaw detection method based on YOLO v2-tone

Abstract

The invention discloses a medical band-aid flaw detection method based on YOLO v2-tone. The methodethod based on YOLO v2-. The method for detecting the defects of the medical band-aid of the tiny comprises the following steps: acquiring image data of the band-aid by using a camera, and making a VOC format data set; T; using the made data set to train a YOLO v2-line model until the loss function isreduced to 0.1-0.2, and storing the weight modelraining YOLO v2-with the made dataset Reducing the loss function to be 0.1-0.2 by the aid of the tiny model, and storing the weight model; returning the specific coordinate position of the band-aid image by using the trained YOLO v2-ty model, and cutting a target band-aid according to the coordinateYOLO v2-USING TRAINED YOLO v2- Returning a specificcoordinate position of the band-aid image by the tiny model, and cutting a target band-aid according to the coordinate; T. Taking the pixel size between 10 * 10 <-30 > * 30 at the center of the cut band-aid, calculating the similarity between the cut target and the white area with the same size by utilizing a single-channel color algorithm, and adopting an Euclidean distance as a similarity calculation formula; a threshold value is set according to the environment, the value range of the threshold value is 3500-5000, if the Euclidean distance calculation result is larger than the value, no flux core exists, and the value smaller than or equal to the value is set to be provided with the flux core. The problem that medical bandages on an existing production line consume a large amount of manpower and financial resources can be effectively solved, and the flaw detection efficiency can reach 40 times per minute.

Description

technical field [0001] The invention belongs to the field of image recognition, and in particular relates to a method for detecting defects of medical band-aids based on YOLO v2-tiny. Background technique [0002] Machine replacement is an important measure to promote industrial upgrading, and target defect detection is one of the important means to improve industrial efficiency. Using machines instead of people to carry out sorting and the ability of machines to identify and classify certain objects will greatly improve industrial efficiency. [0003] In the medical band-aid industry, the defect detection of the band-aid occupies an important position. Improving the accuracy and efficiency of the defect detection of the medical band-aid will greatly improve the production efficiency of the enterprise. The defect detection of medical band-aids is mainly to detect whether there is a drug core. At present, the method of defect detection mainly relies on manual naked eye dete...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the embodiment of the present invention is to provide a medical band-aid defect detection method based on YOLO v2-tiny. The target recognition network combined with the OpenCV color matching algorithm can effectively improve the accuracy and efficiency of defect detection while effectively solving the above-mentioned shortcomings of manual detection.

Images