Model training method, packaging image identification method and device

Abstract

The invention provides a model training method, a packaging image identification method and a packaging image identification device. The model training method comprises the steps of obtaining annotation information corresponding to a first packaging image, wherein the first packaging image is a packaging image of first sample food; analyzing the annotation information to obtain a first text; a target preset text is adopted to at least partially replace the first text to obtain a second text, the target preset text and the first text have an association relationship, and semantics of the second text are description of the second sample food; and training a food classification model according to the first text and the second text to obtain a trained food classification model. According to the embodiment of the invention, the target preset text is adopted to at least partially replace the first text on the first packaging image, so that the expansion of the training sample is realized, the situation that characters on the packaging images of various foods need to be manually labeled to obtain enough training samples is avoided, the labeling workload of workers is reduced, and the working efficiency is improved. And the model training efficiency is improved.

Description

technical field [0001] The present application relates to the field of computer technology, and in particular, to a model training method, a method for recognizing packaging images, and a device thereof. Background technique [0002] Usually, the packaging of food is printed with characters of various shapes, such as italics, word art, etc. Some or all of the textual information in the characters is a description of the properties of the food. It is of great value to accurately extract these textual information describing food attributes. [0003] However, due to the variety of food types and the rich information on the packaging, in order to identify the information on different food packaging, it is necessary to train different food classification models for different foods in advance. At present, when training different food classification models, it is necessary to collect packaging images corresponding to food packaging, and then identify the characters on the packagi...

AI Technical Summary

Problems solved by technology

At present, when training different food classification models, it is necessary to collect the packaging images corresponding to the food packaging, and then use OCR (Optical Character Recognition, Optical Character Recognition) to recognize the characters on the packaging images, and then need the staff to mark the characters that are incorrectly recognized by OCR , and then get accurate training data for the training of the food classification model, which requires manual labeling of a large number of characters on the packaging images, thereby increasing the workload of the staff

Images