Aluminum profile defect detection method based on improved Faster-RCNN

Abstract

The invention discloses an aluminum profile defect detection method based on an improved Faster-RCNN. The method comprises the steps of selecting a plurality of aluminum material defect images, performing data enhancement for a distribution imbalance phenomenon of a data set to set parameters of a Faster-RCNN model; the aluminum profile defect detection method comprising the steps of preprocessingan aluminum profile defect image, training a defect data set on a Faster-RCNN network, storing a final detection model, and detecting aluminum profile defect data by utilizing the trained improved Faster-RCNN model. The improvement points of the method on a Faster-RCNN model are as follows: 1, optimization is performed by adopting a feature pyramid method aiming at an extreme length-width ratio phenomenon of aluminum profile defects; 2, for the phenomenon that the defect shape is irregular, more defect shapes are learned by adopting deformation convolution; 3, for the phenomenon of feature loss of the small target in the feature extraction process, improving the survival rate of target defect features in a high-level network by adopting hole convolution; and 4, aiming at the phenomenon ofinaccurate positioning of small defects, more accurate positioning information is obtained by adopting an ROI Pooling method.

Description

technical field [0001] The invention relates to the fields of computer vision, deep learning and pattern recognition, and in particular to an aluminum profile defect detection method based on an improved Faster-RCNN model. Background technique [0002] As one of the important materials in modern industry, aluminum profiles are widely used in transportation, construction, industrial manufacturing and other fields. Before aluminum profiles are put into use, the quality of aluminum profiles needs to be tested. [0003] The existing aluminum profile quality detection method is to compare products by artificial naked eyes and hand touch feeling. [0004] However, the above methods have at least the following disadvantages: 1. Artificial fatigue is easy to make mistakes in the face of long-term detection; 2. Human judgment standards are subjective, and different judgments will occur for similar samples at different times; 3. Artificial The detection efficiency is very low, and re...

AI Technical Summary

Problems solved by technology

[0005] The present invention proposes a defect target detection method, citing the Faster-RCNN model and adopting various methods to improve the difficult samples in the defect detection process of aluminum profiles, which can solve the problems of poor reliability and low efficiency in manual detection of surface defects question

Images