Integrated scrap steel grade judgment and foreign matter early warning method based on multi-scale image analysis

Abstract

The invention discloses an integrated scrap steel grading and foreign matter early warning method based on multi-scale image analysis, and relates to the technical field of metal treatment. A priori knowledge base is constructed by using overall information of unloading vehicles and a comparative learning method, large-class division is performed on scrap steel images captured in each layer, and a quality inspection process is guided in a targeted manner to call different quality inspection models to perform type identification in combination with priori knowledge and division results. And meanwhile, a cavity convolution strategy is introduced, so that a visual receptive field can be expanded, and an encoder and decoder structure capable of extracting multi-scale information is constructed accordingly. The method comprises the following steps of: performing convolution feature extraction on an image through an FPN network, and enhancing different levels of feature maps by using a repeated attention mechanism based on channel attention and space attention so as to highlight foreign matters and impurity regions in the image;

Description

technical field [0001] The invention relates to the technical field of metal processing, in particular to an integrated scrap steel grading and foreign matter early warning method based on multi-scale image analysis. Background technique [0002] In recent years, a variety of methods have been proposed in academia to improve the accuracy of semantic segmentation and object detection tasks, and atrous convolution is one of the more widely used techniques. In the deep network, atrous convolution can increase the receptive field while maintaining the original resolution and making the target positioning more accurate. This technology is relatively mature in academia, but it is lacking in the industrial field. In addition, in the data collection work of the actual scene, there will be a lot of unutilized unlabeled data. Compared with manually labeled data, unlabeled data has the characteristics of easy availability and large scale, and has higher utilization value. To take ad...

AI Technical Summary

Problems solved by technology

[0003] The existing automated quality inspection and grading systems for scrap steel often only focus on obtaining the grading basis directly from the images, while ignoring the collection, acquisition and utilization of prior knowledge including raw material sources and overall characteristics, which leads to system construction process. In addition, the scale span between different target instances in the semantic segmentation of scrap steel images is large, and the sample scale is not balanced for foreign body early warning and impurity identification. , the amount of data is small; in view of this, we propose an integrated scrap classification and foreign body early warning method based on multi-scale image analysis

Images