Intelligent diamond color grading method

Abstract

The invention discloses an intelligent diamond color grading method, and relates to the technical field of image processing. The method comprises the following steps: step S01, selecting an image preprocessing mode; s02, selecting an assembly; s03, designing an overall structure of the model; s04, selecting a training mode; step S05, carrying out model training; and step S06, carrying out new sample prediction. According to the method, a plurality of images of different angles of the diamond input by a user are preprocessed, a convolutional neural network and a regression device are used as components, a designed model overall structure is used for training a model in a training mode, and an end-to-end model is obtained. A user inputs a plurality of pictures of the diamond at different angles to the model, outputs predicted regression values, compares the regression values with a set threshold value to obtain the color level of the diamond, and replaces manual work to complete the grading task of the diamond color. The diamond color grading efficiency is high, the result is objective and stable, and the diamond color grade can be quickly and accurately given.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to an intelligent method for diamond color grading. Background technique [0002] Image classification is the process of judging which specific category the content of a given image belongs to in an existing fixed set of classification labels. [0003] Viewing angle changes, size changes, lighting conditions, background interference, etc. of the image content are all difficulties in the classification task. In addition, there are many types of diamond colors, subtle features, easy to be affected by light, and difficult to identify. Therefore, it is a relatively difficult task to classify diamond color images. difficult question. Traditional machine learning methods generally first extract image features, and then use feature modeling. The features are generally selected manually, not flexible enough, not accurate enough, and often cannot fully reflect the key information ...

AI Technical Summary

Problems solved by technology

The features are generally selected manually, not flexible enough, not accurate enough, and often cannot fully reflect the key information of the image, so the accuracy rate is not satisfactory

Deep learning technology performs well in the field of computer vision, and can also achieve good results in the diamond color image classification task, but there is still room for further improvement in its accuracy

Images