Junhao Yu

Objects from YCB Dataset Not Visible in Rendered Images Without `--scale 100` When Running `generate_training_data.py`?

2

When using BlenderProc to generate synthetic images of objects from the YCB dataset, the objects do not appear in the rendered images. However, I found that adding the argument `--scale...

It seems that I do not have the permission to download the dataset? 

Hi @paulpanwang , Thank you for your impressive work! However, I was wondering if you could kindly share the pre-trained model weights? I would like to conduct some experiments based...

I know the function is to convert the camera_params predicted in the bounding box image size to original image size, but I didn't understand how `tx, ty` is calcualted? Can...

if the `objaverse-xl` dataset has test descriptions for each object

3

Since `objaverse` dataset have tag or descriptions for each objects, I want to ask if the `objaverse-xl` dataset has descriptions for each object as well. I want to use `objaverse-xl`...

Thanks for your great work! Due to some historical code reason, I couldn't understand how some elements in `base_data` are attained? 1. How the up_sampling and down_sampling matrices for smpl...

Extend VGGT to dynamic reconstruction

1

The VGGT architecture is designed for static scene reconstruction, utilizing its paradigm and multi-loss functions to achieve robust performance in such settings. Given its effectiveness in static environments, can VGGT...

In the current implementation, a third coordinate system, referred to as the Hand-Object Interaction (HOI) coordinate system is introduced to predict the `SDF` of the object before transforming it to...

Why the obman dataset doesn't have `hand_global_rot` annotations?

Hi, I'm working with WebDataset to do test step and my goal is to have a standard training loop where each sample from the dataset is seen exactly once per...