Zheng Chen comments

Results 7 comments of


                                            Zheng Chen

How can I use parameters(cam and pose) to get the camera transform matrix ?

@Miaoing @zhougr18 Any updates to the problem?

how does minimizing the shannon entropy increase the accuracy of classification?

I think minimizing entropy = increasing accuracy makes sense ONLY when the pre-adaptation probability is well-calibrated. But I cannot find any discussion on calibration in the paper...

Organization of the datasets

@Mona77 You need to use the ```g_mask.py``` [here](https://github.com/switchablenorms/CelebAMask-HQ/tree/master/face_parsing/Data_preprocessing) to merge the parts to complete masks. See the instruction [here](https://github.com/switchablenorms/CelebAMask-HQ/tree/master/face_parsing) for more details.

Can't catkin_make

@lericson @wecacuee @w0617 Did you guys successfully run the exploration after training the model?

First frame initialization issues about pose estimation

Hi @HengyiWang, thanks for your explanation. I noticed that in the default code, the `first_frame_mapping` uses the gt pose. Will the results be worse if we do not use the...

Function 'MambaSplitConv1dScanCombinedFnBackward' returned nan values in its 0th output.

@tridao I encountered the same issue in my project. The loss is not nan, but it seems that the grad from one mamba func is nan. I checked the output...

When will the training code released?

hi @zibojia Thanks for bringing this work to the public. I am also very interested in the training process and how you prepared the data. Could you please release a...