TRI-ML/dd3d

Inference Example

ammarali32 opened this issue · 4 comments

ammarali32 commented

Could you please provide a simple inference example?

dennis-park-TRI commented

Hi @ammarali32, does this part of the readme work for you? https://github.com/TRI-ML/dd3d#evaluation

ammarali32 commented

Hi @dennis-park-TRI, Yes it works but it will be really nice if I can test a random video instead of the dataset and see the results directly

dennis-park-TRI commented

Hi @ammarali32, yes, that's do-able. But applying DD3D on a random image needs caution, becuase you also need to know the intrinsics parameter of cameras to correctly determine the size and distance of objects. For now, I will leave this issue open; will work on it soon.

  • 👍4
  • 🚀2
Riccardo240196 commented

Hi @dennis-park-TRI, I am also very interested in applying DD3D online (in a real scenario). Is it possible with the code you provided to extract only the 3D pose estimation part and apply it directly to an image stream? (Of course, the intrinsic parameters of the camera are known).