HawkEyesAI/BAT_Master

🚀 main Branch → Live Deployment

Project: BAT_Master → main

HuggingFace Repo: https://huggingface.co/HawkEyesAI/BAT_Master

⚙️ Full Deployment Command

sudo apt update && sudo apt upgrade -y \
&& sudo apt-get install -y iproute2 libgl1 nano wget unzip nvtop git git-lfs \
&& git config --global credential.helper store \
&& git clone -b main https://huggingface.co/HawkEyesAI/BAT_Master \
&& cd BAT_Master \
&& chmod +x deploy.sh \
&& ./deploy.sh \
&& sleep 10 \
&& python test.py

🔁 Redeploy Script

chmod +x reDeploy.sh && ./reDeploy.sh && sleep 10 && python test.py

🧪 Run API

cd BAT_Master && python BAT_API.py

⚡ Run with Hypercorn

hypercorn BAT_API:app --bind 127.0.0.1:8679 --workers 2

🌐 Expose via ngrok

ngrok http --domain=hasb.nagadpulse.com 8679

🐳 Run with Docker

An alternative to the bare-metal deploy.sh flow above — builds the same FastAPI service into a container instead of installing dependencies directly on the host. See Dockerfile and docker-compose.yml for the full setup.

One-time setup

git lfs pull                        # pulls projectBAT/AI_Models/*.pt (~3GB) — mounted into the container, not baked into the image
cp .env.example .env                # optional: override PORT / PUID / PGID / MEMORY_LIMIT

Build and run — auto-detects GPU vs CPU

./scripts/docker-run.sh up -d --build

scripts/docker-run.sh checks for an NVIDIA GPU + NVIDIA Container Toolkit on this host and layers docker-compose.gpu.yml on automatically when both are present, otherwise it runs the CPU-only docker-compose.yml as-is — no flag to remember either way. It's a plain passthrough to docker compose, so every subcommand below works the same way through it (./scripts/docker-run.sh logs -f, ./scripts/docker-run.sh down, etc.). To pick explicitly instead:

docker compose up -d --build                                                   # force CPU
docker compose -f docker-compose.yml -f docker-compose.gpu.yml up -d --build   # force GPU

Check it's alive

curl http://localhost:8679/BAT-Main/Status/

Logs

./scripts/docker-run.sh logs -f

Stop

./scripts/docker-run.sh down

Test the Docker setup itself

bash scripts/lint-docker.sh          # Dockerfile + compose file lint (hadolint + compose config)
bash scripts/docker-smoke-test.sh    # build, start, wait for healthy, tear down

Publish to Docker Hub

export IMAGE_NAME=<your-dockerhub-username>/bat-master
export IMAGE_TAG=1.0.0
docker login
docker compose build
docker compose push

docker compose build bakes the application source into the image (COPY . . in the Dockerfile) — since this is proprietary HawkEyes AI code, push to a private Docker Hub repository, not a public one. projectBAT/AI_Models/ (git-lfs weights) is excluded from the image via .dockerignore and stays a volume mount — anyone pulling the image separately needs their own git lfs pull of this repo alongside it.

📚 Citation

@misc{hawkeyes_digital_monitoring_ltd_2025,
    author       = { HawkEyes Digital Monitoring Ltd },
    title        = { BAT_Master (Revision c83273f) },
    year         = 2025,
    url          = { https://huggingface.co/HawkEyesAI/BAT_Master },
    doi          = { 10.57967/hf/7062 },
    publisher    = { Hugging Face }
}