Yifan Zhang

Yeah, I agree to what you mean. But what cannot be ignored is that the **sample_points** and **shuffle_points** operation in the data prepossessing, which may affect all the results of...

The memory footprint here is calculated by averaging since our work can process 100 frames in parallel on an RTX2080Ti (11GB total except for the initial memory 800 MB for...

Very sorry for the late reply. Generally speaking, the training setting is always empirical and the result could be infulenced by several factors such as the capability of model as...

Hi, thank you for your attention, I have to apologize for my late response. To be honest, the IoU loss in our work has potential errors in gradient backwards, so...

Is there any progress on this issue?

Hi, @OuyangJunyuan, thanks a lot for your information and deep apologize for late reply. Practically, I found the **QueryDilatedAndGroup** is not powerful than the normal **QueryAndGroup**. However, I will fix...

Sorry for late reply @Divadi , thanks for your attention and it is a very insightful question. The reasons are twofold. The first is that the learning-based sampling strategy is...