zqbai-jeremy / DFNRMVS

Source code for CVPR 2020 paper "Deep Facial Non-Rigid Multi-View Stereo"
GNU General Public License v3.0
167 stars 25 forks source link

The camera parameters #12

Open lin-girl opened 3 years ago

lin-girl commented 3 years ago

How did you get camera parameters information from the dataset?How do I apply for Stirling/ESRC 3D face database?

zqbai-jeremy commented 3 years ago

Sorry for my late reply. The database website is http://pics.stir.ac.uk/ESRC/ We use rendered images for training so the camera parameters are manually set. We only use the textured scans in the database.

lin-girl commented 3 years ago

Thank you. How did you manually set the camera parameters?

zqbai-jeremy commented 3 years ago

For head pose, we randomly sample Euler angles. For pitch and roll, we random sample from a range. For yaw, we random sample a delta angle added on top of previous sample to ensure a roughly uniform view coverage from left side to right side. Then we use sampled head pose to render an image that is relatively large, and use an off-the-shelf face detector to crop this rendered image and get 2D translation and scale.

1-w-m commented 2 years ago

Dear Lin, can you get the camera parameters now?