The HO-Cap Toolkit is a Python package that provides evaluation and visualization tools for the HO-Cap dataset.
HO-Cap: A Capture System and Dataset for 3D Reconstruction and Pose Tracking of Hand-Object Interaction
Jikai Wang, Qifan Zhang, Yu-Wei Chao, Bowen Wen, Xiaohu Guo, Yu Xiang
[ arXiv ] [ Project page ]
- 2024-06-24: The HO-Cap dataset is released! Please check the project page for more details.
If HO-Cap helps your research, please consider citing the following:
@misc{wang2024hocap,
title={HO-Cap: A Capture System and Dataset for 3D Reconstruction and Pose Tracking of Hand-Object Interaction},
author={Jikai Wang and Qifan Zhang and Yu-Wei Chao and Bowen Wen and Xiaohu Guo and Yu Xiang},
year={2024},
eprint={2406.06843},
archivePrefix={arXiv},
primaryClass={id='cs.CV' full_name='Computer Vision and Pattern Recognition' is_active=True alt_name=None in_archive='cs' is_general=False description='Covers image processing, computer vision, pattern recognition, and scene understanding. Roughly includes material in ACM Subject Classes I.2.10, I.4, and I.5.'}
}
HOCap Toolkit is released under the GNU General Public License v3.0.
This code is tested with Python 3.10 and CUDA 11.8 on Ubuntu 20.04. Make sure CUDA 11.8 is installed on your system before running the code.
-
Clone the HO-Cap repository from GitHub.
git clone --rescursive git@github.com:IRVLUTD/HO-Cap.git
-
Change the current directory to the cloned repository.
cd HO-Cap -
Create conda environment
conda create -n hocap-toolkit python=3.10
-
Activate conda environment
conda activate hocap-toolkit
-
Install hocap-toolkit package and dependencies
# Install dependencies python -m pip install --no-cache-dir -r requirements.txt # Build meshsdf_loss bash build.sh # Install hocap-toolkit python -m pip install -e .
-
Download models for external libraries
bash download_models.sh -
Download MANO models and code (
mano_v1_2.zip) from the MANO website and place the extracted.pklfiles underconfig/ManoModelsdirectory. The directory should look like this:./config/ManoModels ├── MANO_LEFT.pkl └── MANO_RIGHT.pkl
-
Run below code to download the whole dataset:
python dataset_downloader.py --subject_id all -
Or you can download the dataset for a specific subject:
python dataset_downloader.py --subject_id 1 -
The downloaded
.zipfiles will be extracted to the./datadirectory. And the directory should look like this:./data ├── calibration ├── models ├── subject_1 │ ├── 20231025_165502 │ ├── ... ├── ... └── subject_9 ├── 20231027_123403 ├── ...
-
Below example shows how to visualize the pose annotations of one frame:
python examples/sequence_pose_viewer.py
-
Below example shows how to visualize sequence by the interactive 3D viewer:
python examples/sequence_3d_viewer.py
-
Below example shows how to offline render the sequence:
python examples/sequence_renderer.py
This will render the color image and segmentation map for all the frames in the sequence. The rendered images will be saved in the
<sequence_folder>/renders/directory.
HO-Cap provides the benchmark evaluation for three tasks:
- Hand Pose Estimation (A2J-Transformer1 and HaMeR2)
- Novel Object Pose Estimation (MegaPose3 and FoundationPose4)
- Novel Object Detection (CNOS5 and GroundingDINO6).
Run below code to download the example evaluation results:
python config/benchmarks/benchmark_downloader.pyIf the evaluation results are saved in the same format, the evaluation codes below can be used to evaluate the results.
-
Evaluate the hand pose estimation performance:
python examples/evaluate_hand_pose.py
You should see the following output:
PCK (0.05) PCK (0.10) PCK (0.15) PCK (0.20) MPJPE (mm) 45.319048 81.247619 91.357143 95.080952 25.657379
-
Evaluate the novel object pose estimation performance:
python examples/evaluate_novel_object_pose.py
You should see the following output:
Object_ID ADD-S_err (cm) ADD_err (cm) ADD-S_AUC (%) ADD_AUC (%) G01_1 0.622285 0.931847 95.251779 93.088153 G01_2 1.722639 2.864552 88.236088 82.951038 G01_3 3.603058 5.267925 80.363333 74.809918 G01_4 3.319628 5.182604 81.892213 73.259688 G02_1 2.876358 4.932917 83.108740 71.551933 G02_2 2.311827 4.164094 85.415819 73.653125 G02_3 2.053942 4.038427 86.666730 73.781861 G02_4 2.156008 4.216609 85.868099 72.308455 G04_1 2.291773 4.423770 84.896350 70.877876 G04_2 2.277173 4.526859 84.796541 69.969442 G04_3 2.262719 4.480607 84.811976 70.126703 G04_4 2.187466 4.335308 85.241063 71.009475 G05_1 2.202152 4.406457 85.158656 70.094175 G05_2 2.150769 4.311178 85.284022 70.394463 G05_3 2.101135 4.209764 85.459741 70.812713 G05_4 2.049368 4.321723 85.748722 69.201963 G07_1 2.239657 4.499831 84.288352 68.425880 G07_2 2.283744 4.585382 84.192769 68.369226 G07_3 2.289358 4.521216 84.392293 69.088029 G07_4 2.453944 4.659746 83.901788 69.095688 G09_1 2.335954 4.383290 84.421006 70.399909 G09_2 2.207153 4.117222 84.960095 71.813927 G09_3 2.335119 4.363489 84.739485 70.545486 G09_4 2.314741 4.390959 84.742636 69.967545 G10_1 2.287382 4.345581 84.872734 70.169253 G10_2 2.292289 4.354261 84.920001 70.067050 G10_3 2.286696 4.332340 84.864717 70.138265 G10_4 2.350560 4.466639 84.312511 69.109810 G11_1 2.478856 4.630755 83.580471 68.318521 G11_2 2.517070 4.716128 83.381718 67.764666 G11_3 2.497892 4.752518 83.509188 67.267398 G11_4 2.608370 4.907855 82.847013 66.485662 G15_1 2.607319 4.912701 82.787732 66.344681 G15_2 2.604308 4.916133 82.790136 66.274095 G15_3 2.603031 4.916675 82.782173 66.238405 G15_4 2.629115 4.932682 82.644975 66.187657 G16_1 2.606751 4.876389 82.686423 66.579694 G16_2 2.583274 4.851990 82.732962 66.555754 G16_3 2.636666 4.903458 82.405020 66.285514 G16_4 2.613952 4.858562 82.467323 66.479288 G18_1 2.623657 4.922163 82.487034 66.229327 G18_2 2.623725 4.909405 82.459508 66.320043 G18_3 2.605120 4.869260 82.583889 66.595389 G18_4 2.582878 4.822793 82.745806 66.909936 G19_1 2.579643 4.815924 82.741131 66.929992 G19_2 2.594446 4.834087 82.630870 66.835297 G19_3 2.589485 4.847906 82.652686 66.650070 G19_4 2.598538 4.853894 82.662542 66.699528 G20_1 2.590124 4.950461 82.710792 65.838859 G20_2 2.572236 4.932026 82.833246 65.916376 G20_3 2.542719 4.877217 83.028086 66.275407 G20_4 2.576188 4.990698 82.730561 65.337352 G21_1 2.563550 4.973498 82.796708 65.404425 G21_2 2.556220 4.961612 82.823936 65.445065 G21_3 2.588855 4.998793 82.592185 65.252844 G21_4 2.608319 5.020533 82.438422 65.145589 G22_1 2.584527 4.989324 82.588827 65.342481 G22_2 2.635756 5.087002 82.387761 64.790779 G22_3 2.643167 5.106887 82.358116 64.694632 G22_4 2.680397 5.162142 82.094643 64.501227 Average 2.680397 5.162142 83.829502 68.882950
-
Evaluate the novel object detection performance:
python examples/evaluate_novel_object_detection.py
You should see the following output: (click to expand)
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.253 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.279 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.248 Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.016 Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.276 Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.249 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.549 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.549 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.549 Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.138 Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.562 Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.388 AP: 0.253 | AP_50: 0.279 | AP_75: 0.248 | AP_s: 0.016 | AP_m: 0.276 | AP_l: 0.249
Footnotes
-
A2J-Transformer: Anchor-to-Joint Transformer Network for 3D Interacting Hand Pose Estimation from a Single RGB Image ↩
-
MegaPose: 6D Pose Estimation of Novel Objects via Render & Compare ↩
-
FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects ↩
-
CNOS: A Strong Baseline for CAD-based Novel Object Segmentation ↩
-
Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection ↩