EcompDetection

Automated Sorting and Defect Detection of Electronic Components Using Computer Vision for PUP CPE Laboratory

EcompDetection is a computer vision-based system designed to assist in the inspection, classification, and sorting of electronic components using a YOLOv8 model. This project is developed for the PUP Computer Engineering Laboratory to aid in the educational and quality assurance processes.

🔍 Features

• Object Detection: Identifies various electronic components and defects using a custom-trained YOLOv8 model.
• Edge TPU Optimization: Includes TensorFlow Lite models compiled for Coral Edge TPU to accelerate inference.
• GUI Application: A full-screen Tkinter-based GUI for real-time detection, object counting, and serial communication with Arduino.
• Automated Sorting: Interfaces with a conveyor or robotic arm system for physical sorting based on detection results.
• Logging and Analysis: Captures detection results and logs them for further analysis.

📁 Project Structure

EcompDetection/
│
├── ecomp_ard/                    # Arduino-related code and hardware interface
├── models/                       # Model files and saved weights
├── py files/                     # Supporting Python scripts and utilities
├── servo-test/                   # Scripts for testing servo motor movements
├── shelf/                        # Optional folder for placing electronics to be detected
├── tflite new/                   # Updated TFLite models (EdgeTPU, etc.)
│
├── ecomp-detect-yolov8n-v1_edgetpu.tflite  # EdgeTPU model (optimized)
├── ecomp-detect-yolov8n-v1.tflite          # Base quantized TFLite model
│
├── inference_ard.py              # Inference with Arduino serial communication
├── inference-gui.py              # Fullscreen GUI application
├── inference.py                  # Inference only (no GUI or serial)
├── README.md                     # Project documentation

🖥️ Requirements

• Python 3.9.12
• OpenCV 4.5.5.62
• PyCoral
• Tkinter
• Ultralytics 8.2.73
• Serial Communication (pyserial)
• Edge TPU Runtime (for Coral USB Accelerator)
• Arduino IDE (for firmware)

🚀 Running the Application

📌 Note: It is recommended to create and activate a virtual environment before running the application to avoid dependency conflicts.

python -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`

Inference test:

python inference.py

With Arduino communication enabled:

python inference_ard.py

For viewing the whole project with GUI:

python inference-gui.py

🤖 Hardware Integration

• Camera: USB webcam (mounted behind or above robotic arm)
• Processing: Raspberry Pi 4B 8GB with Coral USB Accelerator
• Actuator: Servo Motor (controlled via Arduino UNO)
• Components: Resistors, capacitors, LEDs, and defective parts (e.g., rusted, cracked, missing leg)

📌 Status

• [/] Model trained and converted
• [/] Real-time GUI developed
• [/] Edge TPU tested successfully
• [/] Arduino integration done
• Final deployment and enclosure

📜

This project is developed for academic purposes at Polytechnic University of the Philippines and is open for educational and research use.

---

Developed by BSCpE - Group4202