广告图像的生成方法、装置、电子设备和存储介质
By acquiring advertising input information and using a text-based image model to generate advertising images that include text information, the problem of poor performance of Chinese data and poor integration of text and background images is solved. This achieves efficient, personalized, and diverse generation of advertising images and improves visual effects.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING BAIDU NETCOM SCI & TECH CO LTD
- Filing Date
- 2024-11-12
- Publication Date
- 2026-07-17
AI Technical Summary
Existing advertising image generation technologies perform poorly on Chinese data, failing to accurately express user needs, lacking the ability to generate commercial marketing scenarios, and exhibiting poor integration of Chinese copy and background images in advertising images, resulting in a lack of diversity.
By acquiring the input information of the advertisement to be pushed, determining the push prompt words, and using the first text-based image model to generate the target advertisement image including the copy information, the text control capability and image generation quality are enhanced by training with a large language model and a diffusion model, so as to realize the personalization and diversity of the advertisement image.
It improves the efficiency of generating advertising images, enhances the integration of text and background images, meets user needs, increases the visual appeal and diversity of advertising images, and reduces the workload of manually creating templates.
Smart Images
Figure CN119579717B_ABST