README.md

metadata

title: Natsar Demo
emoji: 🚀
colorFrom: red
colorTo: red
sdk: docker
pinned: false
hardware: gpu-t4-small

Introduction

Computer Vision object detection for National Search and Rescue (NATSAR)

Prerequisites

1. Install conda (environment management) using terminal on VScode

   • for mac user [https://www.anaconda.com/docs/getting-started/miniconda/main]
   • for window user [https://www.anaconda.com/docs/getting-started/miniconda/main]

2. Create env from conda by using the following command conda create --name natsar python=3.11 it will create natsar (can be diffent name) environtment for this project. and it is also a good practice to create separate environtment for specific project.

3. Activate the environment conda activate natsar

4. Install PDM (package and dependency manager) to avoid conflict dependency pip install pdm sometimes conda doesn't support some libraries, then pip will be allowed to do. BUT use pip within the natsar env.

5. Intstall packages and dependencies

hello pdm install

Running the project locally

after install dependencies, make sure to activate the environment

1. go to folder src using cd src on terminal
2. run app.py file using pdm run streamlit run app.py

**if cloning from huggingface it might need to mount large file with git lfs use pip install git-lfs then git lfs install then git lfs pull to pull the files to local and pdm run streamlit run app.py to run

Build and Test

• Main app.py file to be placed at root of NATSAR-DEMO repo.
• The app to point to different models that sit within the nominated sub-folders