camera-watch
YOLOv8-based camera surveillance with object detection. Works with any IP camera supporting RTSP streams or HTTP snapshots (Hikvision, Dahua, Reolink, Amcrest, Unifi, and more). Detects 80+ object types (person, car, dog, etc.) and sends notifications with snapshots. Use for motion detection, night watch routines, or security monitoring.
Why use this skill?
Transform your IP cameras into a smart security system. Camera Watch uses YOLOv8 to detect 80+ object types and send instant snapshots to your phone.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/henrikback/camera-watchWhat This Skill Does
Camera Watch is a powerful, AI-driven surveillance solution that integrates directly with your existing IP camera infrastructure. By leveraging the industry-leading YOLOv8 (You Only Look Once) object detection model, the skill monitors real-time video feeds—via RTSP streaming or HTTP snapshots—to identify over 80 distinct object classes. It acts as an intelligent security layer that filters out noise, focusing on what matters: people, vehicles, and pets. When a configured object is detected, the skill automatically generates a snapshot, logs the event, and triggers an immediate notification to your preferred messaging platform. This turns 'dumb' surveillance cameras into an automated security system capable of distinguishing between a delivery person and a neighborhood cat.
Installation
Setting up Camera Watch requires a standard Linux-based environment (or WSL/macOS). First, clone your project directory using mkdir -p ~/camera-watch && cd ~/camera-watch. Initialize a Python virtual environment with python -m venv venv and activate it. Use the provided requirements to install dependencies: pip install opencv-python ultralytics pyyaml requests. Once the environment is ready, move the camera_watch.py script into the directory. You must then configure the config.yaml file to match your network cameras, providing the appropriate RTSP or HTTP credentials, confidence thresholds, and notification settings. Finally, test the setup with python camera_watch.py --test to ensure connectivity before deploying it as a background service.
Use Cases
- Home Security: Receive instant notifications when a person is detected on your front porch or driveway.
- Night Watch Routine: Automate surveillance to start at midnight and shut down at dawn, perfect for monitoring property perimeters while you sleep.
- Package Delivery Tracking: Get alerted as soon as a delivery truck or a person carrying a box enters your camera's field of view.
- Pet Monitoring: Monitor your pets while away, getting notified only when they move into specific areas.
Example Prompts
- 'OpenClaw, start the night watch routine for the front door camera.'
- 'Check the logs to see how many people were detected at the gate today.'
- 'Send me a snapshot from the driveway camera right now.'
Tips & Limitations
To maximize performance, use the yolov8n (Nano) model if you are running on low-powered hardware like a Raspberry Pi or a basic home server; use yolov8s or yolov8m for better precision if your host device has a GPU (NVIDIA Cuda or Apple Silicon). Always configure your confidence threshold between 0.5 and 0.7 to avoid false positives caused by shadows or wind. Keep in mind that heavy object detection is CPU-intensive; if running multiple cameras, ensure your host has sufficient thermal headroom. Regularly clean up the snapshots folder to save disk space, or configure keep_days in your config.yaml to handle automated rotation.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-henrikback-camera-watch": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: network-access, file-write, file-read, external-api, code-execution