HPE-Li-ECCV2024

KO | EN

gitlite — search

#react #ai #llm #android #machine-learning #rust #docker #java #windows #awesome #pytorch #mcp #openai #open-source

HPE-Li-ECCV2024

(★ 12)

This repo is the official implementation for HPE-Li: WiFi-enabled Lightweight Dual Selective Kernel Convolution for Human Pose Estimation (ECCV 2024).

HPE-Li-ECCV2024 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# HPE-Li 
This repo is the official implementation for [**HPE-Li: WiFi-enabled Lightweight Dual Selective Kernel Convolution for Human Pose Estimation**](https://www.ecva.net/papers/eccv_2024/papers_ECCV/papers/04496.pdf), which was published at **ECCV 2024**. And, its extension was published in the TAI journal, named HPE-Li++. This repository was updated for **HPE-Li++** version with better performance than **HPE-Li**, which allows researchers and practitioners to reproduce our results on MM-Fi and WiPose datasets.

## HPE-Li Framework

![image](https://raw.githubusercontent.com/Toandinh1/HPE-Li-ECCV2024/main/Architecture.jpg)

## DSK Convolution

![image](https://raw.githubusercontent.com/Toandinh1/HPE-Li-ECCV2024/main/DSKconv.jpg)

## Data Preparation
### Download datasets.

#### There are 2 datasets to download:

- MM-Fi Dataset
- WiPose Dataset

#### MM-Fi Dataset

1. Request dataset [here](https://ntu-aiot-lab.github.io/mm-fi)
2. Download the WiFi datasets:
   1. `MMFI_Dataset.zip`
   2. `MMFI_action_segments.csv`
3. Unzip all files from `MMFI_Dataset.zip` to `./data/mmfi/dataset` following directory structure:

```
- data/
  - mmfi/
    - dataset/
      - E01/
        - S01/
          - A01/
            - rgb/
            - mmwave/
            - wifi-csi/
              ...
```

#### WiPose Dataset

1. Request dataset [here](https://github.com/NjtechCVLab/Wi-PoseDataset)
2. Unzip all files from `Wi-Pose.rar` to `./data/wipose/` following directory structure:

```
- data/
  - wipose/
    - Train/
    - Test/
              ...
```

#### Person-in-WiFi-3D Dataset (New Dataset)

1. Request dataset [here](https://aiotgroup.github.io/Person-in-WiFi-3D/)
2. Unzip all files from `wifipose_dataset.zip` to `./data/person_in_wifi_3d/` following directory structure:

```
- data/
  - person_in_wifi_3d/
    - train_data/
    - test_data/
```

#### Some new datasets
1. [XRF V2](https://dl.acm.org/doi/10.1145/3749521)
2. [XRF55](https://dl.acm.org/doi/10.1145/3643543)

You guys can read [this survey](https://arxiv.org/pdf/2503.08008v2) to update more interesting things!

## Run Simulations

### MM-Fi dataset
```
python att_mmfi.py 
```

### WiPose dataset

```
python att_wipose.py
```
### Check Complexity

```
python complexity.py
```
### Using Denoiser

```
python denoiser_training.py
```
## Visualization

![image](https://raw.githubusercontent.com/Toandinh1/HPE-Li-ECCV2024/main/Visualization.jpg)

## Acknowledgements

This repo is based on [MetaFi++](https://github.com/pridy999/metafi_pose_estimation) and [SKNet](https://arxiv.org/pdf/1903.06586). The data processing is borrowed from [MM-Fi](https://github.com/ybhbingo/MMFi_dataset) and [WiPose](https://github.com/NjtechCVLab/Wi-PoseDataset).

Thanks to the original authors for their work!

## Citation

Please cite this work if you find it useful.

```
@inproceedings{d2024hpe,
  title={Hpe-li: Wifi-enabled lightweight dual selective kernel convolution for human pose estimation},
  author={D. Gian, Toan and Dac Lai, Tien and Van Luong, Thien and Wong, Kok-Seng and Nguyen, Van-Dinh},
  booktitle={European Conference on Computer Vision},
  pages={93--111},
  year={2024},
  organization={Springer}
}
```

HPE-Li-ECCV2024

HPE-Li-ECCV2024 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

HPE-Li-ECCV2024

HPE-Li-ECCV2024 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README