Créez un scanner du ciel avec IA utilisant Claude

Step 1. Ask Claude to design the basic version of the device. Use this prompt: "Help me design a small AI device that scans the sky with a camera and tracks moving objects in the air." Claude will turn the idea into a simple system with a camera, a small computer, and software that watches the video feed. Step 2. Ask Claude to create the parts list. Use this prompt: "Give me a cheap parts list for an AI sky scanner that uses a camera, Raspberry Pi, Python, OpenCV, and object tracking." The basic build needs a Raspberry Pi, a wide-angle USB camera, a power bank, a tripod, and a simple weatherproof box. Step 3. Ask Claude to write the first camera script. Use this prompt: "Write a Python script that opens a live camera feed, scans the sky for moving objects, and draws a box around anything flying through the frame." This gives you the first working prototype. The device can now watch the air and mark anything that moves through the camera view. Step 4. Ask Claude to add object tracking. Use this prompt: "Add object tracking so every flying object gets its own ID and a movement trail across the screen." Now the device does more than detect movement. It can follow the same object from frame to frame and show where it came from. Step 5. Ask Claude to add object classification. Use this prompt: "Help me classify detected flying objects into simple categories like bird, plane, helicopter, drone, and unknown object." This makes the device feel much smarter. Instead of only showing movement, it tries to understand what type of object is in the air. Step 6. Ask Claude to add a memory system. Use this prompt: "Make the device save a screenshot, timestamp, object type, and movement path every time it detects something in the sky." Now every detection becomes a small record. The device builds a history of what passed through the air and when it appeared. Step 7. Ask Claude to add alerts. Use this prompt: "Make the device send a Telegram alert when it tracks the same flying object for more than three seconds." This turns the prototype into a real monitoring tool. You no longer need to watch the camera feed all day because the device only messages you when something appears. Step 8. Ask Claude to build the dashboard. Use this prompt: "Create a simple web dashboard that shows the live camera feed, object boxes, object IDs, movement trails, timestamps, and saved detections." This makes the project look like a real air scanner. The dashboard shows what the device sees, what it is tracking, and what it has already recorded. Step 9. Put everything into a small box. Mount the camera on a balcony, near a window, or on a tripod with a clear view of the sky. Connect the camera to the Raspberry Pi, run the script, and leave the device scanning the air. The final setup is simple. A camera watches the sky. AI finds moving objects. The tracker follows them. The dashboard records everything. Claude helps you build each part step by step.