2HDED-NET

One defocused image → metric depth map + all-in-focus reconstruction.

 ·  Swagger UI · /info

Input (defocused)
Predicted depth (viridis, 0–10 m)
Predicted all-in-focus