Our take
Inactiveno commits in over a year- Activity
- 20
- Community
- 4
- Issues
- 21
- Pull requests
- 21
+0 stars in the last 8 days
- Fewer contributors than 92% of the projects we track
This is a focused, privacy-oriented prototype, but its year-long lack of commits and single-contributor development make it a higher-maintenance-risk choice for anything safety-critical.
Good fit if
- You can run Python 3.12 or newer, Docker Compose, and a GPU-equipped local inference setup.
- You want to experiment with local video-LLM alerts and can verify the system's behavior yourself.
- You are comfortable adapting MIT-licensed software and maintaining a project with one contributor.
Look elsewhere if
- You need a maintained, safety-certified, or dependable replacement for direct supervision.
- You cannot provide a GPU or operate the Docker Compose, Redis, and vLLM components described by the setup.
- You need evidence of active development or recent issue resolution before adopting a project.
All health signals
| Last commit | 2025-05-22 (1.4 years ago) |
|---|---|
| Commits, last 90 days | 0 |
| Releases, last 12 months | 0 |
| Contributors | 1 |
| Issues closed, last 90 days | 0 |
| Project age | 20 months |
Checked on 2026-10-07 with the GitHub API.
Overview
AI Baby Monitor analyzes webcam or RTSP video against configurable, natural-language safety rules. When its local video-language model judges that a rule may have been broken, it plays a brief beep so a caregiver can check the situation.
It is designed as a privacy-focused extra set of eyes for people comfortable running a local AI stack. The project explicitly describes itself as experimental and not a replacement for adult supervision.
Key Features
- Checks camera frames against rules written in room-specific YAML configuration.
- Uses a local video LLM, with Qwen2.5 VL served through vLLM as the default model.
- Captures video and routes recent and subsampled frames through Redis.
- Plays a single beep when the model returns an alert decision.
- Provides a Streamlit dashboard with the live view and model reasoning logs.
- Supports configuring multiple rooms and camera sources.
- Keeps video and inference within the local network according to the project description.
Use Cases
- A caregiver wants a quiet alert when a configured hazard may occur while they are nearby but briefly distracted.
- A household wants to monitor different rooms using separate camera and rule configurations.
- A developer wants to experiment with local video-language models on a practical camera-monitoring task.
- A privacy-conscious user wants to test video analysis without sending footage to an external service.
What you need
Detected in the repository
- Python >=3.12 (from pyproject.toml)
- A Compose file to run it with its related services
- A test suite
License in plain words
MITpermissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the copyright and license notice
- Share your changes: no
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
git clone https://github.com/zeenolife/ai-baby-monitor.git && cd ai-baby-monitor
cp .env.template .env
docker compose up --build -d
uv run scripts/run_watcher.py --config-file configs/living_room.yaml
See the README for prerequisites, configuration, and dashboard details.
Considerations
The repository is inactive, with no commits or releases in over a year, and most of the code comes from one person. No issues were closed in the recent measured period, so adopters should plan to inspect and maintain the code themselves. Setup also involves a GPU, Python 3.12 or newer, Docker Compose, and local model-serving components; the README notes that the first model download is several gigabytes. Model judgments can miss hazards or raise false alarms, and the project itself warns that it is experimental and must not replace adult supervision.
Found this useful?
Share it with someone who would like ai-baby-monitor.