Training just with RGB and Depth images from LiDAR camera

zubair-irshad / CenterSnap

Pytorch code for ICRA'22 paper: "Single-Shot Multi-Object 3D Shape Reconstruction and Categorical 6D Pose and Size Estimation"

286 stars 47 forks source link

Open alphayama opened 1 year ago

alphayama commented 1 year ago

Hello,

I wanted to know if I can just train the dataset using RGB and Depth images
How can I train model to detect different types of objects other than the ones in the current model?
I was unable to find link to auto_encoder_model ─ model_50_nocs.pth in prepare_data