This project is an attempt to tackle energy wastage in indoor spaces using a combination of computer vision, IoT, and cloud communication.
Instead of keeping lights, fans, or cooling systems running unnecessarily, the system observes how a space is actually being used and makes decisions accordingly.
At a high level, a vision model detects and counts people in real time. This occupancy data is sent to the cloud using MQTT, where it is consumed by an IoT node and a live dashboard. The system then intelligently decides when devices should be turned ON or OFF, factoring in both human presence and environmental conditions such as temperature.
The goal is simple:
use energy only when it is genuinely needed.
-
Computer Vision (YOLO)
Detects and counts people in real time using a camera feed. -
Cloud Communication (EMQX + MQTT)
Enables reliable, low-latency data transfer across different networks. -
IoT Control (ESP32)
Receives occupancy data and controls relays connected to electrical devices. -
Live Dashboard (JavaScript)
Visualizes occupancy, temperature trends, relay states, and estimated energy savings.
Most existing automation systems rely on static schedules or simple motion sensors.
This system instead reacts to actual human behavior, reducing unnecessary power consumption and enabling smarter energy usage.
By combining AI-based perception with IoT control, the project demonstrates a practical and scalable approach to sustainable energy management in classrooms, offices, and shared indoor spaces.
- Real-time vision-based occupancy detection
- Cloud-based MQTT communication (network-agnostic)
- Intelligent relay control using occupancy and temperature logic
- Live analytics dashboard with energy usage estimation
- Modular and scalable architecture
- Python (YOLO, OpenCV)
- ESP32 (WiFi, MQTT, relay control)
- EMQX Cloud (MQTT broker)
- JavaScript (Dashboard & analytics)
- HTML / CSS (UI)
- Smart classrooms
- Office energy optimization
- Shared indoor spaces
- Green building prototypes
- Hackathons and academic projects
This project is built as a proof of concept for intelligent, green energy–aware automation.
It is designed to be extendable with additional sensors, predictive models, and multi-room control logic.