Content-Based Image Retrieval System for Satellite Images Implemented Using Python, Flask And OpenCV.
- Given a query image, this app returns other images from a directory that are similar.
- "Similarity" is either based on the description of an image with a color histogram or with SIFT features. The distance can be calculated using the chi-squared or canberra distance.
-
To use a different image dataset (optional)
- Populate image DB in
app/static/images
- Then in Terminal:
>> python3 -m venv venv >> source venv/bin/activate >> pip install -r requirements.txt >> cd app >> python index.py --dataset static/images --descriptor color --index index_color.csv
- Populate image DB in
-
Run locally using Docker
- Install Docker
- Then in Terminal:
>> docker build --tag=imagesearch . >> docker run -p 80:8000 imagesearch
- You should be able to access app at
localhost:80
in browser
Project is based on the work of Kene Udeh and the original work can be found [here])(https://github.com/fatemaquaid987/Content_Based_Image_Retrieval)