Rice Variety Detection

Project information

  • Category: Python, AI
  • Model Type: Multiple
  • Project Date: 12 March, 2025
  • Project Link: Github

I developed an advanced rice variety detection system using deep learning and computer vision techniques. Leveraging a large dataset of over 75,000 images—each labeled as one of five rice varieties (Arborio, Basmati, Ipsala, Jasmine, and Karacadag)—I experimented with both a transfer learning model and a custom-designed CNN. The custom CNN achieved exceptional performance with a test accuracy exceeding 99.5%, while the transfer learning model also demonstrated robust results. This project, implemented using TensorFlow and deployed on Kaggle, highlights my ability to build and fine-tune complex image classification pipelines for practical applications in agricultural quality control and food supply management.