booker-max

QuestionsAbout the acquisition of SMPL parameters in the Human3.6M data set

2

I’m sorry to bother you again. I saw a sentence in your paper "Alternatively, we obtain groundtruth SMPL parameters by applying SMPLify-X on the groundtruth 3D joint coordinates of Human3.6M...

Questions about testing on human3.6M

3

Hello, I would like to ask if the pose_coord_gt_h36m you provided on the Human3.6M test set was obtained by using Smplify-x to estimate the SMPL parameters? 

Regarding the calculation of the three indicators The PA MPVPE, PA MPJPE, and F-scores when test on the Freihand dataset

2

Hello, when I tested the network on the Freihand dataset, the annotation in the testset did not provide joint_cam, and mano parameters, your code uses dummy data, and uses the...

Some thoughts on improving the model

4

1.Hello, improving the resolution of the heatmap is helpful to the improvement of the indicators of MPJPE and PA-MPJPE. Have you tried to further increase the resolution of the heatmap...

Is there a problem with the pre-trained model snapshot_12.pth.tar?

4

 I used snapshot_12.pth.tar for demo, but the effect is not very good. Is anyone in the same situation as me?

Hello, can you provide a few samples of the event dataset you generated based on the REDS dataset, I would like to debug create_hdf5.py

17

Your work is really great, it gave me a lot of inspiration, but I still have some parts that I did not understand, especially how to obtain the derivative_{gt}, I...

hello, the pretrained models of flow is not accesible, can you upload again. Thank you.

Your article is a great inspiration to me, can you provide the training code, I want to do further research.

您好，在tab.2中， zerodce的性能是20.06db/0.61， 我想问一下这个结果是怎么得到的呢，因为这是一个无监督的方法， 您是给它加入了gt的约束来训练么？ 还是 直接使用作者提供的预训练模型？ 亦或是在使用zerodce作者提供的预训练模型基础上在sdsd数据集上重新训一遍等等。 并且还想问一下，tab.2中给出的实验结果是在训练集125个视频片段(包括室内和室外)训练，在测试集25个视频片段(包括室内和室外)测试得到的么？ 因为我自己将zerodce在sdsd数据集上进行训练之后，发现得到的结果和您表格中的出入较大，所以想要了解一下里面的细节，谢谢。