Hi, thanks for sharing this great work. When you train the model you using only 2D images from different view or a combination of 2D+3D images ? Thank you