Repository Issue Activity (beta)

#484 Low conf for far away objects?

#482 How can I use enhance_depth.py for general images

#480 About the fine-tuning train_loss_conf_depth hits a plateau

#481 About custom training VGGT from scratch

#369 VGGT Finetuning Issue on VKITTI Datasets

#479 About mixing ratio for synthesized data and real-world data？

#466 Visualized Network Attention Layer

#470 I found that the cause of the OOM (Out of Memory) issue in demo_colmap is due to the resize dimension problem.

#476 Training VGGT from scratch

#411 Can VGGT fine-tune on datasets with cameras，without depth？

#106 VGGT on large datasets

#475 Using panoramic images to train the VGGT model, the predicted depth map has all values being a constant distribution with very low variance.

#421 How to initialize VGGT with known camera exirinsics?

#436 Multiplicated shifted geometry

#424 Will there be an accumulation of errors in the predicted camera poses?

#417 Scale Issues in Model Normalised Space

#456 qvec issue always there when run the -use- ba

#455 you can reduce cuda mem usage by discarding intermediate tensors

#407 Applicability of VGGT for Long Sequence Images

#425 Applying texture to 3D point map output

#461 How to evaluate the pose information output by VGGT

#471 Are VGGT output depth maps metric or relative?

#192 Long video input for visual odometry

#254 pose errors are large compare to sfm or slam

#312 Comparison of 3DGS Training Results: Output from demo_colmap.py performs worse than COLMAP's output in training