zhezh/totalcapture-toolbox

A toolbox for processing Total Capture dataset

totalcapture-toolbox's Issues

dataset

hello！ when i enter the https://cvssp.org/data/totalcapture/data/ i can't find the calibration.cal file ! can you help me

Shouldn't gendb.py and conifg.yaml be in the main directory?

I say this because when I tried to run gendb.py from gendata/gendb.py, it couldn't load the tools modules. So I moved gendb.py and config one directory up in the tree and it worked.

can not find pose

there is running pose in your config.yaml. Where can I find running?

thx!

TotalCapture Distortion

Hi, I wonder how you dealt with the distortion of cameras.

The given distortion parameter is too small, so using 1st order radial distortion, that parameter seems like the system is undistorted at all.

I wonder if you suffered the same issue, and if yes, how did you deal with that? Doesn't it affect a lot while reconstructing 3D pose from given 2D points?

Thank you very much in advance!

Error in gendata/gendb.py File

Hi,
I have followed the instruction to process total capture dataset you have described. After organizing all the fills as you have mentioned, while I run the command, python gendata/gendb.py it was running perfectly and giving verbose like the followings:
100%|███████████████████████████████████████| 3385/3385 [00:45<00:00, 74.29it/s]
100%|███████████████████████████████████████| 3385/3385 [00:45<00:00, 74.34it/s]
100%|███████████████████████████████████████| 2421/2421 [00:32<00:00, 74.43it/s]

But unfortunately after a long time it gives a key error as bellow:
Traceback (most recent call last): File "./TotalCapture-Toolbox-master/gendata/gendb.py", line 368, in <module> train_dataset = extract_db(config, train_dir_meta, cameras) File "./TotalCapture-Toolbox-master/gendata/gendb.py", line 129, in extract_db vid_frame_num = int(vid_info['video']['@nb_frames']) KeyError: 'video'
Can you please help me resolve it?

IMU data

Does this repo also help in extracting IMU data from the dataset? If yes, which section of the code is working on IMU data.

which two coordinate systems are being transformed by R_{Ti}

Hello, I'm currently encountering an issue with rotation matrices when manually inputting measurement data.

I'd like to inquire about which two coordinate systems are being transformed by R_{Ti}?
From the section (3.4) implementation details of the paper "Real-Time Full-Body Motion Capture from Video and IMUs" referenced by the authors of the TotalCapture dataset, I understand that R_{ig} is assumed to be consistent between IMUs and aligned with the world coordinates through the global up direction and magnetic north.
Despite this, I'm still unable to grasp which two coordinate systems RTi represents the transformation matrix for. Therefore, I'd like to ask for clarification from everyone. Thank you!

about dataset where to download

│ │ ├── acting1_Xsens_AuxFields.sensors
│ │ ├── gt_skel_gbl_ori.txt
│ │ ├── gt_skel_gbl_pos.txt
│ │ ├── s1_acting1_calib_imu_bone.txt
│ │ ├── s1_acting1_calib_imu_ref.txt
│ │ ├── s1_acting1_Xsens.sensors

where can i find these files, I know acting1_BlenderZXY_YmZ.bvh file is in Vicon Groundtruth section, but where are the rest files, and gt_skel_gbl_pos.txt I really cannot find it.

关于数据对齐

您好，想问下您的处理过程中有包括IMU，Vicon和视频数据时间对齐的处理么

Total capture supports SMPL generation?

Thanks for the work!

I came here through a paper "Animatable NeRF" for reconstruction. I am seeking to generate SMPL values for Humam3.6m using TotalCapture.
Could you inform which data I wokkd require and the structure. I have the files for humam3.6m with me, just need to generate accurate smpl joints. ( if possible with the groundtruths )

Can you provide any insights on this matter if the SMPL is possible via Total Capture,

Thank you