Core Features (4-Part System)
- Fruit Freshness Detection (Image Classification + ESP32)
- Trained an ML model to classify Fresh vs Not Fresh
- Real-time implementation using:
- ESP32
- DHT22 sensor
- Dashboard options:
- Capture new photos OR select from a folder
- Product Detection + IR Conveyor Counting (YOLO)
- YOLO model detects product on live camera
- IR sensor + ESP32 counts items on a conveyor
- Result: Real-time product tracking automation
- OCR: Expiry Date / MRP / Manufacturing Details Extraction
- YOLO detects the printed region
- Code crops & extracts:
- → MRP
- → Expiry date
- → Manufacturing details
- OCR: Brand & Product Name Extraction
- YOLO locates brand area
- OCR extracts text
- CSV maps brand to database (Product identification)
Components Used
- Hardware
- ESP32 microcontroller
- IR sensor (for conveyor product counting)
- DHT22 temperature & humidity sensor (fruit freshness logic)
- Conveyor belt mechanism (DIY setup)
- USB/Webcam for live product detection
- AI / Machine Learning
- YOLOv5 / YOLOv8 (for product detection + region detection)
- Roboflow (dataset labeling & dataset preparation)
- OCR / Text Extraction
- EasyOCR / Tesseract OCR
- CSV mapping for brand & product identification
- Software / Development
- Python (AI/ML + OCR scripts)
- Jupyter Notebook (model training & testing)
- VS Code (project folder & YOLO environment)
- Arduino IDE (to program ESP32)
Working
Working
Code & Installation
The complete source code is available in the GitHub repository.
My Learnings
- Building ML pipelines (dataset → training → deployment)
- YOLO dataset creation, training, and inference
- OCR text extraction + region cropping using ML detection
- Integrating AI + IoT (ESP32 + IR sensors)
- Real warehouse automation workflow:
- Detect → Identify → Extract → Count
