/augier-prediction-model

Primary LanguageJupyter Notebook

AI Based Image Retrieval

Instructions for deployment on GPU machine or server:

  • Clone the repository
  • Create conda environment from environment.yml file
  • Go to --> deployment_latest --> run app.py and the file will run on the local server

Deployment Instructions

Github

Local Machine Setup

conda env create -f environment.yml

Confirm if the environment with the name ‘pytorch’ is created:

conda env list

If the environment is present then activate the environment: conda activate pytorch

  • cd deployment_latest
  • Run command: python app.py The application will run on the local server and will now respond to postman requests.

Server Machine Setup

Login to the server using the .pem file. Place the .pem file in a folder and open the terminal in that folder. Run the following commands one by one: chmod 400 Image_retrieval.pem

ssh -i “Image_retrieval.pem” ubuntu@ec2-35-175-222-98.compute-1.amazonawas.com

You will be logged into the ec2 instance. Now you have to git clone the repository: https://github.com/mkhb654/augier-prediction-model.git

cd augier-prediction-model

Create conda environment from the .yml file in the folder:

conda env create -f environment.yml

Confirm if the environment with the name ‘pytorch’ is created:

conda env list

If the environment is present then activate the environment: conda activate pytorch

cd deployment_latest
Run command: python app.py

The application will run on the server and will now respond to postman requests.

Code files

The app.py (→  aurgier-prediction-model → deployment_latest → app.py ) file is the api file. It can opened in a text editor like VS code on a local machine and can be opened on the server machine using the command: nano app.py