# CIGPose **Repository Path**: Tomhappy/CIGPose ## Basic Information - **Project Name**: CIGPose - **Description**: No description available - **Primary Language**: Unknown - **License**: Apache-2.0 - **Default Branch**: main - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2026-09-21 - **Last Updated**: 2026-09-21 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README

CIGPose: Causal Intervention Graph Neural Network for Whole-Body Pose Estimation

Bohao Li | Zhicheng Cao | Huixian Li | Yangming Guo

Paper (arXiv:2603.09418)

## Abstract State-of-the-art whole-body pose estimators often lack robustness, producing anatomically implausible predictions in challenging scenes. We posit this failure stems from spurious correlations learned from visual context, a problem we formalize using a Structural Causal Model (SCM). The SCM identifies visual context as a confounder that creates a non-causal backdoor path, corrupting the model's reasoning. We introduce the **C**ausal **I**ntervention **G**raph **Pose** (CIGPose) framework to address this by approximating the true causal effect between visual evidence and pose. The core of CIGPose is a novel Causal Intervention Module: it first identifies confounded keypoint representations via predictive uncertainty and then replaces them with learned, context-invariant canonical embeddings. These deconfounded embeddings are processed by a hierarchical graph neural network that reasons over the human skeleton at both local and global semantic levels to enforce anatomical plausibility. Extensive experiments show CIGPose achieves a new state-of-the-art on COCO-WholeBody. Notably, our CIGPose-x model achieves 67.0% AP, surpassing prior methods that rely on extra training data. With the additional UBody dataset, CIGPose-x is further boosted to 67.5% AP, demonstrating superior robustness and data efficiency. ## 📄 Table of Contents - [📖 Introduction](#introduction) - [🖼️ Visualizations](#visualizations) - [📊 Model Zoo & Results](#model-zoo--results) - [🛠️ Installation](#installation) - [🏃 Usage](#usage) - [🙏 Acknowledgements](#acknowledgements) - [📄 License](#license) - [📚 Citation](#citation) ## 📖 Introduction CIGPose is a whole-body pose estimation framework that improves robustness in challenging scenes (e.g., occlusion, clutter, difficult lighting) by explicitly addressing **visual confounding** via a causal perspective. It is implemented as an [MMPose](https://github.com/open-mmlab/mmpose) project under `mmpose/projects/cigpose`. - **Causal formulation**: Model visual context as a confounder and target the interventional distribution `P(Y|do(F))` instead of the observational `P(Y|F)`. - **Causal Intervention Module (CIM)**: Use predictive uncertainty to identify confounded keypoint embeddings and replace them with learned, context-invariant **canonical embeddings**. - **Hierarchical graph reasoning**: Perform local (intra-part) and global (inter-part) message passing on deconfounded embeddings to enforce anatomical plausibility. ## 🖼️ Visualizations ### 1) Overall results + typical failure case of confounding

- **(a) Quantitative comparison** on COCO-WholeBody val: CIGPose achieves strong accuracy while remaining data-efficient. - **(b) Qualitative comparison** vs. RTMPose-x: baseline predictions may latch onto spurious background cues; CIGPose mitigates this by intervening on confounded keypoint representations, producing more anatomically plausible poses. ### 2) Qualitative comparison on challenging scenes

- **From left to right**: input image, RTMPose-x, CIGPose-x. - CIGPose is more robust under common confounders (e.g., occlusion and clutter), yielding cleaner and more coherent whole-body structures. ### 3) More qualitative results

- Additional examples further illustrating the robustness gains and improved anatomical consistency brought by causal intervention + hierarchical graph reasoning. ## 📊 Model Zoo & Results #### Results on COCO-WholeBody v1.0 val | Config | Input Size | FLOPS (G) | Body AP | Foot AP | Face AP | Hand AP | Whole AP | ckpt | | ------------------------------------------------------ | ---------- | --------- | ------- | ------- | ------- | ------- | -------- | ------------------------------------------ | | [CIGPose-m](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-m_8xb64-420e_coco-wholebody-256x192.py) | 256x192 | 2.3 | 69.0 | 64.3 | 82.1 | 49.7 | 59.9 | [pth](https://drive.google.com/file/d/1GvTAyJGUIVI-5KWKRumqx1YH4dEjmwBv/view?usp=sharing) | | [CIGPose-l](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-l_8xb64-420e_coco-wholebody-256x192.py) | 256x192 | 4.6 | 71.2 | 69.0 | 83.3 | 54.0 | 62.6 | [pth](https://drive.google.com/file/d/1mC5GkU7v_M3tIi8NaDaKDrYvyde-ca2J/view?usp=sharing) | | [CIGPose-l](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-l_8xb32-420e_coco-wholebody-384x288.py) | 384x288 | 10.7 | 73.0 | 72.0 | 88.3 | 59.8 | 66.3 | [pth](https://drive.google.com/file/d/1q_j2bA3A5UubBDEeyHo2z1MPrcDSOMSt/view?usp=sharing) | | [CIGPose-x](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-x_8xb32-420e_coco-wholebody-384x288.py) | 384x288 | 18.7 | 73.5 | 72.3 | 88.1 | 60.2 | 67.0 | [pth](https://drive.google.com/file/d/1S08T9Mvpqjt9TGtQ1Tc-5X8HRDff157d/view?usp=sharing) | | [CIGPose-l+UBody](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-l_8xb32-420e_coco-ubody-256x192.py) | 256x192 | 4.6 | 71.3 | 66.2 | 83.4 | 55.5 | 63.1 | [pth](https://drive.google.com/file/d/1KtgCcbIPIC1EGyIKjXvVF6aag1ow_9xY/view?usp=sharing) | | [CIGPose-l+UBody](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-l_8xb32-420e_coco-ubody-384x288.py) | 384x288 | 10.7 | 73.1 | 72.3 | 88.0 | 61.2 | 66.9 | [pth](https://drive.google.com/file/d/1HAhF_tZNOxY4hNWTKooSnMGAguK1U8Pu/view?usp=sharing) | | [CIGPose-x+UBody](mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-x_8xb32-420e_coco-ubody-384x288.py) | 384x288 | 18.7 | 73.5 | 70.3 | 88.4 | 62.6 | 67.5 | [pth](https://drive.google.com/file/d/13RmDZpWku876gsrrMso6DfnK7F4Z4vZ2/view?usp=sharing) | #### Results on COCO val2017 | Config | Input Size | FLOPS (G) | Params (M) | AP | AR | ckpt | | ------------------------------------------------------------ | ---------- | --------- | ------- | ------- | ------- | ------------------------------------------------------------ | | [CIGPose-m](mmpose/projects/cigpose/body_2d_keypoint/cigpose-m_1xb384-420e_coco-256x192.py) | 256x192 | 1.9 | 14 | 76.6 | 79.3 | [pth](https://drive.google.com/file/d/1tNXidCKVhXqxT8WIYPEdSfIEownbIl5E/view?usp=sharing) | | [CIGPose-l](mmpose/projects/cigpose/body_2d_keypoint/cigpose-l_1xb256-420e_coco-256x192.py) | 256x192 | 4.2 | 28 | 77.6 | 80.3 | [pth](https://drive.google.com/file/d/1jsKXa4waJKkLFmF7IhWVA_xMBs5_61zr/view?usp=sharing) | | [CIGPose-l](mmpose/projects/cigpose/body_2d_keypoint/cigpose-l_1xb64-420e_coco-384x288.py) | 384x288 | 9.4 | 29 | 78.5 | 81.1 | [pth](https://drive.google.com/file/d/14GUos6gnw3Rm78SPPIb48DnK5gzhzFAb/view?usp=sharing) | #### Results on CrowdPose test set | Config | Input Size | Params (M) | AP | AP easy | AP medium | AP hard | ckpt | | ------------------------------------------------------------ | ---------- | ---------- | ---- | ------- | --------- | ------- | ------------------------------------------------------------ | | [CIGPose-m](mmpose/projects/cigpose/body_2d_keypoint/cigpose-m_1xb64-210e_crowdpose-256x192.py) | 256x192 | 14.4 | 71.4 | 81.0 | 72.7 | 58.9 | [pth](https://drive.google.com/file/d/1nH_ONU-4CbaUs6xh20Jqhj_gHf_HFtmi/view?usp=sharing) | | [CIGPose-l](mmpose/projects/cigpose/body_2d_keypoint/cigpose-l_1xb64-210e_crowdpose-256x192.py) | 256x192 | 28.4 | 73.7 | 82.8 | 75.1 | 61.2 | [pth](https://drive.google.com/file/d/1_q2GlrZxOvtK2aQLx79G-B7h3s3kPLgM/view?usp=sharing) | | [CIGPose-l](mmpose/projects/cigpose/body_2d_keypoint/cigpose-l_1xb64-210e_crowdpose-384x288.py) | 384x288 | 28.8 | 74.2 | 82.9 | 75.6 | 62.5 | [pth](https://drive.google.com/file/d/10DYfrnQstJZ3WjWY3I9d9R7cxlptlrNL/view?usp=sharing) | | [CIGPose-x](mmpose/projects/cigpose/body_2d_keypoint/cigpose-x_1xb64-210e_crowdpose-384x288.py) | 384x288 | 50.4 | 75.8 | 84.2 | 77.3 | 63.6 | [pth](https://drive.google.com/file/d/1VN_OhFUeKtwvm0Pdyj5mt5mLwlVCIXaO/view?usp=sharing) | ------ ## 🛠️ Installation Our code is based on [MMPose](https://github.com/open-mmlab/mmpose). ``` # 1. Create a conda environment conda create -n cigpose python=3.8 -y conda activate cigpose # 2. Install PyTorch (adjust CUDA version as needed) pip install torch==2.0.1+cu118 torchvision==0.15.2+cu118 -f https://download.pytorch.org/whl/cu118/torch_stable.html # 3. Install MMCV and MMDetection pip install -U openmim mim install mmengine mim install "mmcv>=2.0.1" mim install "mmdet>=3.1.0" # 4. Clone this repository git clone https://github.com/53mins/CIGPose.git cd CIGPose/mmpose pip install -v -e . ``` ------ ## 🏃 Usage #### Data Preparation Please refer to MMPose guidelines for dataset preparation: - [COCO-WholeBody](https://github.com/open-mmlab/mmpose/blob/main/docs/en/dataset_zoo/2d_wholebody_keypoint.md) - [COCO](https://github.com/open-mmlab/mmpose/blob/main/docs/en/dataset_zoo/2d_body_keypoint.md) - [CrowdPose](https://github.com/open-mmlab/mmpose/blob/main/docs/en/dataset_zoo/2d_body_keypoint.md%23crowdpose) ## ⭐Train a model ``` cd mmpose bash tools/dist_train.sh mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-l_8xb32-420e_coco-wholebody-384x288.py 8 ``` ## ⭐Test a model ``` bash tools/dist_test.sh mmpose/projects/cigpose/wholebody_2d_keypoint/cigpose-l_8xb32-420e_coco-wholebody-384x288.py path/to/checkpoint.pth ``` ------ ## 🙏 Acknowledgements - This project is built on top of [MMPose](https://github.com/open-mmlab/mmpose), and follows its training/testing utilities and dataset conventions. ## 📄 License This project is released under the [LICENSE](LICENSE). ## 📚 Citation If you find CIGPose useful in your research, please consider citing: ```bibtex @article{li2026cigpose, title={CIGPose: Causal Intervention Graph Neural Network for Whole-Body Pose Estimation}, author={Li, Bohao and Cao, Zhicheng and Li, Huixian and Guo, Yangming}, journal={arXiv preprint arXiv:2603.09418}, year={2026} } ```