Ruicheng Wang

Hi. I have tried a few strategies with pytorch built-in ONNX tools but failed unfortunately. I am not familiar with ONNX and don't know how to fix it. I am...

Hi, thanks for your interest in our work, and apologies for the late response 🙏 The training data must have scale-invariant depth (i.e., rooted at 0, **with no unknown shift**)...

Really appreciate your continued interest! We will release the training code approximately by the time of the paper's availability, when we are back from CVPR 🏃.

Hi! The glb follows the convention of OpenGL identity camera coordinate system, +x right, +y up, +z backward. (The raw output point map follows the OpenCV convention, +x right, +y...

Hi, thank you for your interest and for raising these valuable questions! 1. I agree that IRLS can be a viable alternative, particularly as a simpler compromise for the L1...

Great idea! I agree that recovering shift with known intrinsics can be very useful in many scenarios. Although the model itself is not aware of intrinsics input, the output point...

Hi, thanks a lot for your contribution! However, the current PyTorch implementation already supports native CPU inference. Could you share the specific reasons or advantages you see for re-implementing part...

Hi, thank you for trying out our model and providing feedback. The holes in the visualized depth map are due to MoGe's mask prediction. MoGe includes an additional head to...

Hi, thanks for your interest! We wished to open-source the data refinement code if possible. However, the pipeline depends on a model trained on synthetic data. In our experiments, this...

@CongliangLi @jackchinor Here are some details that might be helpful: * **Depth filtering in the first stage is critical.** It's important to remove misalignments in the ground-truth depth as thoroughly...