Prompt word template generation method and device based on artificial intelligence, equipment and medium

By building a target image database and optimizing the insertion position of the descriptive word, a prompt word template is generated, which solves the problem of frequent iteration of propt in generative AI, and quickly generates character pictures of specific skin colors, improving efficiency.

CN120337893APending Publication Date: 2025-07-18SHENZHEN PINGAN COMM TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510495655.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-04-18
Publication Date
2025-07-18

Smart Images

  • Figure CN120337893A_ABST
    Figure CN120337893A_ABST
Patent Text Reader

Abstract

The invention relates to the field of artificial intelligence and financial science and technology, and discloses a cue word template generation method and device based on artificial intelligence, equipment and a medium, and the method comprises the steps: obtaining a target picture containing a target physiological feature, and constructing a target picture database; generating a picture generation prompt word corresponding to the target picture through a preset picture generation text model, extracting a description word describing the target physiological feature, and constructing an initial description word library; and inputting the description words in the initial description word bank into a text graph model to generate a plurality of first text pictures, testing the accuracy rate of the target physiological features of the first text pictures through a classification model, screening out the description words with the accuracy rate higher than a preset threshold value, and constructing a target description word bank. And respectively inserting the target description words into different positions of the preset cue words, generating multiple groups of second text pictures through a text picture model, testing the accuracy rate of the second text pictures through a classification model, selecting the insertion position with the highest accuracy rate as the target insertion position, generating the cue word template, avoiding frequent iteration of the cue words, and realizing rapid generation.
Need to check novelty before this filing date? Find Prior Art