Eyesight – Machine Vision for Enhancing Quality and Production Efficiency

Computer vision-based quality inspection system developed as part of Google Bangkit 2024 Capstone Project, designed for UMKM manufacturing sector.

Eyesight: Machine Vision for Enhancing Quality and Production Efficiency

Overview

Eyesight is a computer vision-based quality inspection system developed as part of the Google Bangkit 2024 capstone project.
The system is designed specifically for micro and small enterprises (UMKM) in the manufacturing sector, aiming to provide an accessible and cost-effective solution for defect detection.
Utilizing YOLOv8 for object detection and deployed on the Jetson Nano, the system operates entirely offline to support areas with limited connectivity.

Contribution

  • Machine Learning Lead:
    • Collected and labeled visual data relevant to UMKM products.
    • Trained a custom YOLOv8 model using Ultralytics and optimized it for real-time inference with TensorRT.
    • Deployed the model on Jetson Nano for edge-based inspection.
    • Collaborated with cloud and mobile teams for system integration and UI.

Results

  • Successfully demonstrated real-time product inspection on edge devices.
  • Increased awareness of AI-driven quality control in small-scale industries.
  • Received positive feedback for its potential to support local manufacturers through automation and efficiency improvements.
  • Recognized by Bangkit reviewers as an initiative with scalable social impact.

Documentation