README.md

Server

Prerequisite

1. Set up elasticsearch and kibana.

docker-compose -f ./elasticsearch/docker-compose.yml up -d

2. Create a .env file same folder as the server.py, you can use this sample

ES_HOST=elasticsearch
ES_PORT=9200
ES_INDEX=flickr-images

3. Create flickr-images index in elasticsearch

python create_index_es.py

Local set up

1. Install all the necessary libraries

pip install -r requirements.txt

2. Create the necessary model artifact (if not already created)

python ../model/clip_model.py

3. Run the server

ES_HOST=localhost uvicorn server:app --host 0.0.0.0 --port 80

Docker container set up

1. Build the docker image, run this from parent directory

docker build -t image-search-server -f ./server/Dockerfile .

2. Run the container

docker run --name image-search-server --network elasticsearch_default -p 80:80 image-search-server

The api can be accessed at localhost:80