👶 Baby Alert Detection System

A Raspberry Pi–powered real-time monitoring system that detects the presence of a baby in a locked car using an AI-powered camera model. If a baby is detected, the system sends an instant alert via Telegram.

---

🧠 Project Overview

• Goal: Prevent life-threatening scenarios by alerting parents if a baby is detected in a locked car.
• Technology: Raspberry Pi, USB camera, YOLOv11 model, Telegram Bot integration.
• Detection Flow:
  1. Lamp ON = car is unlocked.
  2. Once the lock button is pressed, the system activates the camera and model.
  3. If a baby is detected, a Telegram alert is sent to the parent.

---

🛠 Hardware Components

• 🧠 Raspberry Pi – central controller.
• 📷 USB Camera – for capturing real-time video.
• 💡 LED Lamp – indicates whether the car is locked or unlocked.
• 🔘 Button – simulates the car locking mechanism.

---

💻 Software Stack

• Python – programming language used throughout the project.
• OpenCV – handles video streaming and frame processing.
• PyTorch + YOLOv11 (Ultralytics) – for object detection.
• RPi.GPIO – controls the Raspberry Pi GPIO pins.
• Label Studio – used for annotating training images.
• Telegram Bot API – used to send alerts when a baby is detected.

---

🧪 Custom Model Training Process

Collect and Label Data & Train the Detection Model

• We collected 500 images of babies from various sources.
• Training is handled via Train_YOLO_Models.ipynb.
• YOLOv11 from Ultralytics was trained on our custom dataset.
• The trained model is exported and moved to the Raspberry Pi for deployment.

---

📲 Telegram Bot Integration

• Telegram bot created with BotFather.
• Alerts sent via Telegram Bot API using Python.