The application is described in further detail in two blog posts (part 1 and part 2).
Web application build using FastAPI containing 3 endpoints:
- Landing page with simple interface for uploading '.csv' file for determining DVPI score (Danish stream plant index, link to report in Danish) or images for species identification
- Species identification using deep learning model trained to identify 100 plant species included in DVPI
- Determine DVPI score by calling an external SOAP API endpoint using a '.csv' file with plant species and cover (example file)
Repository contains all necessary file to run the application. To install packages for runtime environment:
pip install -r requirements.txt
To run the application locally run:
uvicorn main:app --reload
The repository also contains additionally files for serving the application on Heroku.
Landing page interface