Apple 与加州大学研究人员合作,发布了带有自然语言指令的开源人工智能图像编辑模型 MGIE。 Apple releases open-source, AI-powered image editing model MGIE with natural language instructions, collaborating with UC researchers.
苹果与加州大学的研究人员合作推出了新的人工智能图像编辑模型 MGIE。 Apple has launched a new AI image editing model, MGIE, in collaboration with researchers from the University of California. MGIE 代表 MLLM 引导图像编辑,使用多模态大语言模型 (MLLM),允许用户根据自然语言指令编辑图像。 MGIE, which stands for MLLM-Guided Image Editing, uses multimodal large language models (MLLMs) and allows users to edit images based on natural language instructions. 该模型在 2024 年国际学习表示会议上的一篇论文中提出,展示了其在保持竞争性推理效率的同时改进自动指标和人工评估的能力。 The model was presented in a paper at the International Conference on Learning Representations 2024, showcasing its ability to improve automatic metrics and human evaluation while maintaining competitive inference efficiency.