ameybhavsar24 / sbert-image-search

Search similar images using SentenceTransformers on a interface built with Flask.

sbert-image-search

Do you know how Google's Image search actually works?

I built an image search app that lets you search from 25,000 nature-themed photos from Unsplash.

The application uses SentenceTransformers, a Python framework for state-of-the-art sentence, text and image embeddings. It provides models that allow embedding images and text into the same vector space. This allows us to find similar photos and implement image search.

This app uses Flask as the server and TailwindCSS for frontend components.

Installation

1. Clone this repository.

git clone https://github.com/ameybhavsar24/sbert-image-search.git

2. Create a python virtual environment and activate it.

python -m venv env
(for powershell users)
.\env\Scripts\Activate.ps1

3. Install the project dependencies from requirements.txt.

pip install -r requirements.txt

4. Download the required models (pickle files) from Google Drive in a /models folder in the root directory. You should have the following folder structure for models:

sbert-image-search/
    - models/
        - image_embeddings.pkl
        - image_names.pkl
        - text_to_image_cpu.model.pkl

5. Run the project, it should start an server at 127.0.0.1:5000.

python app.py