Picture type identification method and device

Abstract

The invention provides a picture type identification method and apparatus. The method comprises the following steps: inputting a to-be-classified picture into a pre-trained target detection model to output an identification result, wherein the identification result comprises the type and number of to-be-identified target objects in the to-be-classified picture; adding a corresponding preset score to the score variable every time a to-be-identified target object of a corresponding type appears in the to-be-classified picture, wherein the initial value of the score variable is a preset initial score; and identifying the type of the to-be-classified picture according to the score result of the score variable, wherein the intersection of score results of different types of to-be-classified pictures is zero. According to the picture type identification method and device provided by the invention, the types and the number of the to-be-identified target objects in the to-be-classified pictures are identified by utilizing the target detection model, the score variables are processed according to the types and the number of the to-be-identified target objects to obtain the score result, and the types of the to-be-classified pictures are obtained according to the score result. The picture classification efficiency and accuracy are improved, and the robustness of the classification model is improved.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method and device for identifying a picture type. Background technique [0002] Various application scenarios will have the need to judge the type of the picture according to the situation of the target object in the picture. The traditional method usually consists of various types of pictures to form training samples for training to obtain an image classification model, and classify the pictures according to the image classification model. But usually, the classification accuracy of the image classification model is low due to the little difference between pictures of different classifications. [0003] For example, in judging the compliance of rental housing, it is necessary to judge the number and type of beds in the indoor image uploaded by the user. If the image does not contain such beds, it is classified as 0, and if it contains one bed or two beds, it is clas...

AI Technical Summary

Problems solved by technology

[0004] The traditional method uses the method of image classification model to judge the category, but if there are few pictures of the two categories, it will lead to serious uneven distribution of training data and little difference between images of different categories, which will also cause great damage to the accurate classification of the image classification model. difficulty

Images