Detection method, computer equipment and storage medium

Abstract

The embodiment of the invention discloses a detection method. The method comprises the following steps of inputting the object picture into a backbone network of a defect detection network for featureextraction; obtaining feature data, inputting the feature data into a plurality of branch networks of the defect detection network, depending on feature data, respectively detecting different types of flaws on the object picture by a plurality of branch networks; and outputting defect detection results. The method is advantaged in that each branch of the neural network can be used for specifically and independently detecting one type of flaws; a problem that the detection precision is low when the neural network learns multiple defect forms at the same time is avoided, on the other hand, lesscomputing resources are consumed compared with the mode that multiple neural networks are adopted for detection, and defect detection precision and efficiency are improved under the condition that the same computing resources are consumed.

Description

technical field [0001] The present application relates to the technical field of data processing, and in particular, to a detection method, a computer device, and a computer-readable storage medium. Background technique [0002] With the continuous upgrading of industrial automation, higher and higher requirements are put forward for the application of artificial intelligence in the industrial field. In the field of industrial image flaw detection, there are many types of flaws, and the morphological differences between different flaws are uneven. This feature of industrial defects is a major problem faced by defect detection and classification. [0003] As we all know, defects of different forms have different effects on business, so customers have high requirements for the accuracy of defect classification. If the defect category is wrongly classified, it will cause two losses: First, the large-impact defect is incorrectly detected. Small-impact defects will cause unexpe...

AI Technical Summary

Problems solved by technology

[0003] As we all know, different forms of defects have different impacts on business, so customers have relatively high requirements for the accuracy of defect classification. If the defect categories are misclassified, it will cause two losses: First, the cost of misdetection of large-impact defects is Small-impact defects will cause unexpected serious problems during product use, causing serious losses to downstream customers and causing complaints from downstream customers; second, small-impact defects are misdetected into large-impact defects, resulting in product pricing Low, sell the product at a low price, resulting in loss of customer income

[0004] The applicant has found through research that when the detection object has many blemish forms and the morphological differences between different morphological blemishes are uneven, the distinction between the blemishes with large differences can be achieved relatively well, but the blemishes with similar morphological differences It is prone to classification errors, and there are problems of low detection accuracy and low detection efficiency

Images