Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/tucan9389/image-clustering-browser
https://github.com/tucan9389/image-clustering-browser
clustering faiss flask k-means-clustering react viewer vision
Last synced: 9 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/tucan9389/image-clustering-browser
- Owner: tucan9389
- License: apache-2.0
- Created: 2020-12-24T07:22:20.000Z (almost 4 years ago)
- Default Branch: main
- Last Pushed: 2020-12-28T08:21:18.000Z (almost 4 years ago)
- Last Synced: 2024-10-07T07:21:04.955Z (about 1 month ago)
- Topics: clustering, faiss, flask, k-means-clustering, react, viewer, vision
- Language: Python
- Homepage:
- Size: 252 KB
- Stars: 3
- Watchers: 3
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Image Clustering Browser
You can browse clustered images on web. This repo also provides clustering with with [faiss](https://github.com/facebookresearch/faiss) similarity and [k-means](https://en.wikipedia.org/wiki/K-means_clustering) clustering algorithm.
## DEMO
## Features
- Server (python 3.6+, flask)
- [x] Serve hierarchical directories browsing api
- [x] Serve images serving api
- Client (javscript, react)
- [x] View the result of clustering with huge number of images
- [x] Browse the hierarchical directories
- [x] Show bunch of images with grid gallery view and paginization
- Others
- [x] Without copying images, cluster million level images and write at txt files (faiss)
- [ ] Make featuremaps with images
- [ ] Run clustering with buttons and show the progress of the task on the web
- [ ] Show the featuremaps for each image
## Diagram
## Getting Started
### 0. Clone the project
```shell script
$ git clone https://github.com/tucan9389/image-clustering-browser
cd image-clustering-browser
```
### 1. Install Requirements
1. [Install anaconda](https://docs.anaconda.com/anaconda/install/) and pathon 3.6+
3. Install python libraries for clustering batch and api server
```shell script
$ pip install numpy Flask Flask-Cors
```
4. [Install yarn](https://classic.yarnpkg.com/en/docs/install/) for react app
5. yarn install on browser-react directory
```shell script
$ cd browser-react
$ yarn install
```
### 2. Prepare Images and Clustering Result
You can select one of the options:
- Download sample data from [image data(19.31GB)](https://drive.google.com/drive/folders/1qY3i-JaR0txMU4UIuVJ65WKUCBQASEL6?usp=sharing) and [clustering data(1.3MB)](https://drive.google.com/drive/folders/1gHwcpuV8bp8-6PeNlyY5yazGIgY0SU7w?usp=sharing)
- Or make your own clustering with [CLUSTERING_GUIDE.md](CLUSTERING_GUIDE.md)
The clustering makes the directory structure. If you make cluster by yourself, you have to conform the structure.
```
../data/clutering-results
└── my-images-001-clustering
├── img_names.txt
├── 20201225-160650-d999-c5
| ├── 000
| | └── img_names.txt
| ├── 001
| | └── img_names.txt
| ├── 002
| | └── img_names.txt
| ├── 004
| | └── img_names.txt
| └── etc
| └── img_names.txt
└── 20201225-170423-d999-c3
├── 000
| └── img_names.txt
├── 001
| └── img_names.txt
| └── 20201226-034123-d103-c3
| ├── 000
| | └── img_names.txt
| ├── 001
| | └── img_names.txt
| └── 002
| └── img_names.txt
└── 002
└── img_names.txt
```
### 3. Run Browser API and Client Server
And then, run the server and client app.
```shell script
# run api server
$ cd browser-server
$ python app.py \
--img_dir_path "../data/my-images-001" \
--clustering_output_path "../data/clutering-results/my-images-001-clustering"
# and then run client server
$ cd ../browser-react
$ yarn start
```
Now you can browse the clustered images at [`http://localhost:3000/browser/`](`http://localhost:3000/browser/`).
## Used Libraries
- [React](https://reactjs.org/)
- [Flask](https://palletsprojects.com/p/flask/)