MIGAN: {
bytes: 28079181;
id: "migan-pipeline-v2";
licence: "MIT";
model: {
bytes: 28079181;
sha256: "6f1f3530a1a2324b19752018ce756088b07973cda8d7d890034ace5c8a48c40b";
url: "https://huggingface.co/andraniksargsyan/migan/resolve/406830d0fa60666da0071c342ad2fbc8f30c5c64/migan_pipeline_v2.onnx";
};
source: "https://huggingface.co/andraniksargsyan/migan";
} = ...
Type Declaration
-
Readonlybytes: 28079181
-
Readonlyid: "migan-pipeline-v2"
-
Readonlylicence: "MIT"
-
Readonlymodel: {
bytes: 28079181;
sha256: "6f1f3530a1a2324b19752018ce756088b07973cda8d7d890034ace5c8a48c40b";
url: "https://huggingface.co/andraniksargsyan/migan/resolve/406830d0fa60666da0071c342ad2fbc8f30c5c64/migan_pipeline_v2.onnx";
}
-
Readonlysource: "https://huggingface.co/andraniksargsyan/migan"
MI-GAN (Picsart AI Research, ICCV 2023; code and weights MIT): object removal. The "pipeline" ONNX export takes an 8-bit image and mask of any size and handles resizing and blending itself.