continue revolution
continue revolution
这应该不是，我也不知道原因，需要GroundingDINO开发人员的帮助。
@longjingling 你可能没有下载cuda，你得去nvidia官网下载跟pytorch对应版本的cuda，你看看你的pytorch是cu117还是什么版本的，版本相同才可以编译。
遵循以下步骤： 1. nvidia官网下载匹配cuda 2. 环境变量查看CUDA_PATH 3. 启动本地命令行（需要是你使用webui的环境）然后跑以下代码 ``` from torch.utils.cpp_extension import CUDA_HOME print(CUDA_HOME) ``` 看看有没有cuda环境 4. 卸载groundingdino，要么pip uninstall groundingdino，要么自己手动从虚拟环境删除文件夹 5. 重新启动webui
我写的插件实现了这个功能，但是背景是透明的 见https://github.com/continue-revolution/sd-webui-segment-anything
你想要白色背景的话，需要去源代码里面找一下np.array([0, 0, 0, 0]) 把它改成白色的RGBA。 你也可以自己在PS里加背景。
Thank you for your highlight! I am currently working on integrating your work into my extension. Stay tuned!
I have done supporting Grounded-SAM and merged it to the master branch. @rentainhe
> The inpainting model may need more considerations in implementation and perhaps we just get other models first. I’m the author of [sd-webui-segment-anything](https://github.com/continue-revolution/sd-webui-segment-anything) and I am planning to connect my...
> > > The inpainting model may need more considerations in implementation and perhaps we just get other models first. > > > > > > I’m the author of...