Data Engine

A complete data annotation and tracking system with both Python (FastAPI) backend and C++ interface. Supports video object segmentation using SAM 2 and point tracking using CoTracker.

Use git clone --recursive https://github.com/Rumarino-Team/data-engine.git when cloning the repo

Install Dependencies

1. Create Python virtual environment using venv or Anaconda (mainly written and tested on Python 3.12.12)
2. If you didn't use --recursive when doing git clone, then run git submodule update --init --recursive to make sure sam2 is downloaded
3. Run pip install -e . while in sam2 directory
4. Run pip install -e ".[notebooks]" while in sam2 directory
5. Return to the data engine directory and run pip install -r requirements.txt

Using the Backend

1. source the venv or Anaconda (depending on your shell)
2. Run fastapi dev backend/api.py to use the API manually with hot reload, Run fastapi run backend/api.py to use the static API

Add the --host xxx.xxx.xxx.xxx and/or --port xxxx args to change its ip and port

Testing the Python Backend

1. Run python3 backend/tests/tester.py to test the API automatically and see results under backend/tests/

Angular Web Frontend

Mainly for developing the frontend, for production its recommended to use the Tauri Desktop env instead

1. Change into frontend-ng/data-engine
2. Install dependencies with bun install or npm install
3. Start the frontend with bun run start or npm run start
4. Keep the Python backend running separately

The frontend defaults to http://127.0.0.1:8000 for backend requests and also exposes a compact API URL override in the top bar.

Tauri Desktop Frontend

1. Change into frontend-ng/data-engine
2. Install JavaScript dependencies with bun install or npm install
3. Install the Rust toolchain and Tauri prerequisites for your OS
4. Start the Python backend separately
5. Run bun run tauri:dev or npm run tauri:dev:npm for the desktop dev workflow
6. Run bun run tauri:build or npm run tauri:build:npm to produce a packaged desktop build

The Tauri app only wraps the Angular frontend. It does not bundle or launch the Python backend.

Testing via C++

Currently this part is behind the python api's backend implementation so it isnt recommended to use

1. Install nlohmann_json and curl dev packages via your package manager
2. Change into /cpp-backend-bindings
3. Run mkdir build
4. Run cmake build to generate CMake build files
5. Run cmake --build build to compile project
6. Run ./build/src/BackendInterfaceTests while the venv is sourced to test backend code

Structure

TBA