苹果公司与加州大学圣塔巴巴拉分校(UCSB)的研究人员共同推出了一款名为MGIE的图片编辑框架,该框架利用多模态大模型MLLM解决了指令引导不足的问题。MGIE通过端到端训练,使扩散模型能同步更新,从而执行图像编辑任务。在人类指令的引导下,MGIE可以进行类似Photoshop的图像修改、全局优化和局部对象修改。这一创新技术有望为图片编辑领域带来革命性的变革。
英文标题:Apple and UCSB Release Open-Source Image Editing Framework MGIE
英文关键词:Apple, UCSB, MGIE, Image Editing, Open Source
英文新闻内容:
Researchers from Apple and the University of California, Santa Barbara (UCSB) have jointly developed the MGIE image editing framework, which addresses the issue of insufficient instruction guidance through the use of the multimodal large model MLLM. By end-to-end training, the diffusion model can be synchronously updated, enabling image editing tasks to be executed. Under human instruction, MGIE can perform image modifications similar to Photoshop, global optimization, and local object editing. This innovative technology is expected to bring about a revolutionary change in the field of image editing.
【来源】https://www.jiqizhixin.com/articles/2024-02-05-10
Views: 1