A wine label recognition method, comprising:
- obtaining a wine image, and performing optical character recognition (OCR) on the wine image in a preset OCR manner, to obtain text comprised in the wine image;
- performing deep learning recognition on the wine image in a preset deep learning recognition manner, to obtain an image feature comprised in the wine image;
- and
- sifting out a target wine label matching the text and the image feature from a preset wine label database according to the text and the image feature, and using the target wine label as a wine label corresponding to the wine image, comprising:
- counting a text similarity between the text and preset wine label text corresponding to an existing preset wine label in the preset wine label database;
- determining whether the text similarity is greater than or equal to a preset text similarity threshold;
- using the preset wine label text as target text, and using the target text as text corresponding to the wine image in a case where the text similarity is greater than or equal to the preset text similarity threshold;
- sifting out preset wine labels matching the text from the preset wine label database according to the text, and combining the preset wine labels into a collection to obtain a target wine label collection;
- sifting out a wine label with a highest matching degree with the image feature from the target wine label collection according to the image feature as the target wine label, and using the target wine label as a wine label corresponding to the wine image;
- wherein the wine image comprises a plurality of images of the same target wine;
- using a target wine label corresponding to each image as a wine label corresponding to the image;
- classifying wine labels belonging to the same wine label among all the wine labels respectively corresponding to the images into the same type of wine labels;
- counting the number of wine labels comprised in each same type of wine labels;
- and
- selecting, according to the number of wine labels comprised in each same type of wine labels, a wine label with the largest value among all the same types of wine labels as a target wine label corresponding to the target wine;
- wherein the step of sifting out a wine label with a highest matching degree with the image feature from the target wine label collection according to the image feature as the target wine label, and using the target wine label as a wine label corresponding to the wine image comprises:
- counting an image similarity between the image feature and a preset wine label included in the target wine label collection to obtain all the image similarities;
- sifting out a target image similarity with the highest image similarity among all the image similarities;
- determining whether the target image similarity is greater than or equal to a preset image similarity threshold;
- and
- using a preset wine label corresponding to the target image similarity as the target wine label, and using the target wine label as the wine label corresponding to the wine image in a case where the target image similarity is greater than or equal to the preset image similarity threshold.