Note: This model has been trained for approximately 2.7M steps (batch size = 1) and is still in the training process. I have attached a .ipynb file in the repository. You can refer to it to know how ...
Abstract: Tracking and reconstructing deformable objects with little texture is challenging due to the lack of features. Here we introduce “invisible markers” for accurate and robust correspondence ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback