Surface Defect Detection Algorithm for Glazed Ceramic Tiles Based on Improved Faster RCNN

In terms of surface defect detection of glazed tiles, there are problems such as false detection and insufficient ability to identify small target defects due to the rich texture on the surface of glazed tiles. In response to these challenges, this article has compiled a dataset of glazed ceramic tiles and proposed an improved algorithm ECANet-FPG based on Faster R-CNN. Due to the similarity in texture and defects on the surface of ceramic tiles, this algorithm introduces the Efficient Channel Attention (ECA) module into the feature extraction network of Faster R-CNN to enhance the ability to extract deep features. At the same time, a Feature Pyramid Grids (FPG) structure is adopted to achieve efficient fusion of multi-scale features, in order to improve the recognition accuracy of small target defects. The experimental results show that the improved algorithm achieves higher detection accuracy and efficiency than the baseline model, demonstrating its effectiveness and practicality in detecting surface defects on glazed ceramic tiles.