LookoutConnect is a professional-grade edge computing middleware designed to bridge standard IP cameras with the LookOut AI Detection Platform. It automates the process of capturing imagery, performing AI analysis, and triggering a multi-layered alert system.

The system is built on a "Hub and Spoke" model. The main controller (Hub) manages the detection logic and system integrity, while specialized Modules handle external services.

• lookoutConnect.py: The central controller. Manages timers, image capture, AI uploads, and process locking.
• pushover.py: Dispatches high-priority mobile alerts with image attachments.
• ntfy.sh: Provides a secondary, low-latency notification channel.
• openWeather.py: Fetches real-time localized weather data for environmental context.
• acsPro.py: Interfaces with Axis Camera Station Pro to trigger local pop-up messages, camera liveviews, or alarms.

• Network Resilience: Uses exponential backoff for uploads to handle unstable remote links.
• System Integrity: A Watchdog Timer monitors every execution cycle. If a network socket hangs, the script hard-terminates to allow a fresh start.
• Cross-Platform Support: Uses specialized file-locking (msvcrt for Windows, fcntl for Linux) to prevent redundant processes.
• Forensic Logging: Self-rotating logs and an image archive provide a permanent audit trail of all AI detections.
• Smart Scheduling: The --runs argument allows the script to run for a specific number of 1-minute duty cycles.

Run the provided install.sh script to set up the directory structure and install dependencies:

chmod +x install.sh
./install.sh

Update the .env file with your camera credentials and API keys. This is the single source of truth for the entire system.

API Mode (Pull): Direct snapshot from the camera URL.

python3 lookoutConnect.py api --runs 60

FTP Mode (Push): Monitor a local directory for new images.

python3 lookoutConnect.py ftp --resize

1. Capture: Image is pulled from the camera or detected in the FTP inbox.
2. Analyze: Image is optionally resized and uploaded to LookOut AI Detection Platform.
3. Enrich: If a wildfire (or other supported object) is detected, localized weather data is fetched.
4. Alert:

• Pushover: Emergency alarm sounds on mobile devices.
• ntfy.sh: Real-time push notification sent.
• ACS Pro: Local physical alarm/relay is activated.

5. Archive: The evidence image and AI metadata are stored for forensic review.

To ensure your system is working correctly before a fire actually occurs, LookoutConnect includes a manual test mode. This allows you to bypass the camera network and "push" a specific local image through the entire AI analysis and notification pipeline.

The --file argument is used specifically with FTP mode. It tells the script: "Ignore the camera folder; process this specific file instead."

Find a representative image (preferably one containing smoke or a controlled burn for a positive test) and place it in your script folder. Let's call it test_fire.jpg.

Run the following command in your terminal or command prompt:

python3 lookoutConnect.py ftp --file test_fire.jpg

When you run this command, the script performs the following steps:

1. Bypasses State Tracking: It ignores the .ptr file, so you can test the same image multiple times.
2. AI Analysis: It uploads test_fire.jpg to the LookOut platform.
3. Full Alert Chain: If the AI detects smoke in your test image, it will trigger the Pushover alarm, ntfy message, and the ACS Pro rule just as it would during a real event.
4. Logging: The results will be recorded in detectionResults.txt with a "Manual" tag.

When testing with a local file, verify the following:

• Pushover: Does the notification include the "Emergency" repeating sound?
• Weather: Does the message correctly display the current temperature from the openWeather module?
• Logs: Check archive/detectionResults.txt to see if the AI confidence scores are being recorded correctly.