Canny Edge Filtered VGG Framework for Diamond Cut Classification

Precise diamond cut categorization is essential for value estimation and uniform grading in the field of diamond quality evaluation. Conventional techniques depend on heuristic algorithms and manual examination, which can be error-prone and time-consuming. In this work, we provide a novel method for diamond cut classification by proposing Canny Edge Filtered VGG16 (CEF-VGG16) that has been improved by preprocessing the diamond images with Canny edge recognition. When preprocessing diamond images, the Canny edge filter is used to highlight the edges and structural characteristics that are essential for differentiating various cut levels. The Diamond Cut Dataset having 1600 images with 8 classes of diamond cuts was used for this execution. The Original images are subjected to extract sobel Edge filtered, Otsu threshold Edge filtered and Canny Edge filtered diamond images. These edge filtered images are applied to existing CNN models. The suggested approach improves resistance to changes in illumination and image quality while simultaneously simplifying the categorization process. The experiment reveals that the proposed CEF-VGG16 model performs well in the diamond cut categorization with a high accuracy of 99.72%.