Image scene recognition method and apparatus

A computing device, comprising:

• a memory and a processor, wherein the memory is configured to store computer-executable instructions, and wherein the computer-executable instructions upon execution by the processor cause the processor to implement operations comprising:
• obtaining a to-be-recognized image;
• inputting the to-be-recognized image into a first sub-model of a machine learning model;
• recognizing a plurality of visual elements in the to-be-recognized image by the first sub-model of the machine learning model;
• determining related visual elements among the plurality of visual elements by performing semantic analysis on the plurality of visual elements;
• clustering the related visual elements to obtain at least one target visual element, each of the at least one target visual element carries an element probability;
• encoding the at least one target visual element in a preset encoding manner and generating an encoding vector of the at least one target visual element, wherein encoding the at least one target visual element and generating the encoding vector of the at least one target visual element comprises:
• determining a vector length of the encoding vector based on an element quantity of preset visual elements, and
• determining an encoding value at each encoding position in the encoding vector having the vector length based on the at least one target visual element and the element probability corresponding to each of the at least one target visual element;
• inputting the encoding vector of the at least one target visual element into a second sub-model of the machine learning model;
• and
• generating a scene category corresponding to the to-be-recognized image by the second sub-model of the machine learning model.

A non-transitory computer-readable storage medium, storing computer-executable instructions, wherein when the computer-executable instructions are executed by a processor, the computer-executable instructions cause the processor to implement operations comprising:

• obtaining a to-be-recognized image;
• inputting the to-be-recognized image into a first sub-model of a machine learning model;
• recognizing a plurality of visual elements in the to-be-recognized image by the first sub-model of the machine learning model;
• determining related visual elements among the plurality of visual elements by performing semantic analysis on the plurality of visual elements;
• clustering the related visual elements to obtain at least one target visual element, each of the at least one target visual element carries an element probability;
• encoding the at least one target visual element in a preset encoding manner and generating an encoding vector of the at least one target visual element, wherein encoding the at least one target visual element and generating the encoding vector of the at least one target visual element comprises:
• determining a vector length of the encoding vector based on an element quantity of preset visual elements, and
• determining an encoding value at each encoding position in the encoding vector having the vector length based on the at least one target visual element and the element probability corresponding to each of the at least one target visual element;
• inputting the encoding vector of the at least one target visual element into a second sub-model of the machine learning model;
• and
• generating a scene category corresponding to the to-be-recognized image by the second sub-model of the machine learning model.