https://github.com/runningleon/albumcluster
https://github.com/runningleon/albumcluster
Last synced: 9 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/runningleon/albumcluster
- Owner: RunningLeon
- Created: 2020-12-28T11:26:10.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2021-01-08T19:27:16.000Z (over 5 years ago)
- Last Synced: 2025-02-08T10:44:04.822Z (over 1 year ago)
- Language: Python
- Size: 4.55 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Album Clustering
## Brief Introduction
This repository contains a tool to cluster albums by face.
The main procedures are:
- Input an album, which has photos of multiple people.
- Detect all face bounding boxes and five facial landmarks using [RetinaFace](https://github.com/deepinsight/insightface/tree/master/detection/RetinaFace).
- Align all cropped face images using five facial landmarks.
- Extract all 512-D face features using [insightface](https://github.com/deepinsight/insightface).
- Cluster all face features with [Chinese Whispers](https://en.wikipedia.org/wiki/Chinese_Whispers_(clustering_method)) and output a clustered album.
## How to use this tool?
### Clone Repository
```bash
git clone git@github.com:RunningLeon/AlbumCluster.git
export ALBUM_HOME=$(pwd)/AlbumCluster
```
### Create Environment
Strongly suggest use `conda` to manage Python environment.
If you do not have conda on your system, please goto download and install [miniconda](https://docs.conda.io/en/latest/miniconda.html)
- create conda environment
Once `conda` is installed, please follow steps below to create the environment.
```bash
conda create -n env python==3.7.0
source activate env
pip install -r requirements.txt
```
- Install mxnet-gpu
By defaults, there is mxnet-cpu in [requirements.txt](./requirements.txt). If you have cuda installed on your system, you can install mxnet-gpu. Supported version is from 1.2 to 1.6.
First confirm the cuda version on your system, than use `pip` to install.
Example for cuda==10.2
```bash
pip install mxnet-cu102==1.6.0
```
- Compile rcnn
```bash
cd $ALBUM_HOME/AlbumCluster/app/retinaface
make
```
- Download models
Link: [Baidu YunPan](https://pan.baidu.com/s/1SQHu5fHB8SGB-7bJ0WmErQ)
Code: `fsfl`
Download `models.zip` and put it in `$AlbumCluster/models` directory.
```bash
cd $AlbumCluster/models
unzip models.zip
```
### Run
```bash
python run.py --input ./images --output ./output
```
After a few minutes, results would be saved to `output` directory.
## Example Results
## FAQs
- None
## References
- [RetinaFace](https://github.com/deepinsight/insightface/tree/master/detection/RetinaFace)
- [insightface](https://github.com/deepinsight/insightface)
- [Chinese Whispers](https://blog.csdn.net/u011808673/article/details/78644485/)