Wildlife Detection API

A REST API that identifies animals in images using Microsoft MegaDetector v5a, built with Flask and PyTorch.

Setup

1. Create a virtual environment

python -m venv venv
source venv/bin/activate  # macOS/Linux

2. Install dependencies

pip install -r requirements.txt

3. Download model weights

python app/download_model.py

This downloads the MegaDetector v5a model (~268MB) to weights/md_v5a.0.0.pt.

Usage

Start the server

python -m app.app

The API runs on http://localhost:5000 by default.

Detect animals in an image

curl -X POST -F "image=@path/to/photo.jpg" http://localhost:5000/predict

Response:

{
  "animal": "animal",
  "confidence": 0.95
}

API Reference

Endpoint	Method	Description
/predict	POST	Upload an image (multipart form, key: image) to detect animals

Error responses:

• 400 — No image provided or empty filename
• 500 — Server/model error

Running Tests

python -m pytest tests/

Or with unittest:

python -m unittest discover tests/

Project Structure

wildlifeApp/
  app/
    app.py              # Flask API entry point
    model_handler.py    # MegaDetector inference logic
    download_model.py   # Model weight downloader
  tests/
    test_images/        # Test fixtures
    test_model_handler.py
  weights/              # Model weights (not committed)
  requirements.txt

Tech Stack

• Framework: Flask
• ML Model: Microsoft MegaDetector v5a (YOLOv5)
• Runtime: PyTorch + torchvision
• Image Processing: Pillow