Abstract: With the in-depth application of information technology in the field of industrial manufacturing, the research of big data in industrial manufacturing is becoming an important reference basis for realizing intelligent manufacturing and helping the government guide the transformation and upgrading of manufacturing enterprises.In the traditional steel, aluminum and other metal manufacturing industry, there are problems such as extensive production mode and simple production process.Therefore, it is urgent to use the new generation of information technology such as artificial intelligence to improve the production process and improve production efficiency.When using aluminum, the surface must be inspected.The existing aluminum surface defect detection is limited by the traditional manual visual inspection, very laborious, or based on the traditional machine vision algorithm, the recognition rate is not high, usually can not accurately determine the surface defects in a timely manner.To solve these problems, Convolutional Neural Networks and YOLOv3 are used to detect the aluminum defect data set produced by two target detection algorithms.Then, based on YOLOv3 algorithm, improvements were made to improve the detection effect of small defects on aluminum surface.Experimental verification was carried out on the "aluminum profile defect recognition" data set provided by Guangdong Industrial Intelligence Big Data Innovation Competition. The experimental results showed that the improved algorithm's mean Average Precision (mAP) was 3.4% higher than YOLOv3 algorithm and 1.8% higher than Faster R-CNN algorithm.
