Fine-grained image classification method, system and device and storage medium

Abstract

The invention discloses a fine-grained image classification method, system and device, and a storage medium. The method comprises the following steps: obtaining to-be-classified image information; andinputting the image information into an identification model trained by adopting a local area to carry out fine-grained identification, and outputting an identification classification result. According to the method, the local area is adopted to carry out enhanced model training, so that the model is more suitable for fine-grained identification, and when two commodities are highly similar, the identification model can more effectively grasp key distinguishing information in the commodity images, so that the commodities can be rapidly and accurately identified and classified, and the method can be widely applied to the field of image data processing.

Description

technical field [0001] The invention relates to the field of image data processing, in particular to a fine-grained image classification method, system, device and storage medium. Background technique [0002] Commodity recognition technology based on image recognition and target detection algorithms has great application potential in new retail and unmanned stores. For example: Through product identification, customers can automatically record when they choose products, and with mobile payment methods, it is possible to pay without queuing at the exit, thereby improving the operational efficiency of supermarkets. However, general-purpose target detection algorithms focus more on detection based on content such as shape, texture, and color. For common commodity packaging, in the distinction of various commodities under the same category, the regularity of its appearance is often very strong (often are cuboids, cylinders, etc.), which can only be distinguished by relying on ...

AI Technical Summary

Problems solved by technology

However, general-purpose target detection algorithms focus more on detection based on shapes, textures, colors, etc., and for common commodity packaging, in the distinction of various commodities under the same category, the regularity of their appearance is often very strong (often are cuboids, cylinders, etc.) and can only rely on local content differences to differentiate, so these algorithms tend not to perform well

Images