CropAI is an AI-powered application developed to explore rice variety identification using image classification and video analysis.
The project combines computer vision, deep learning and a Streamlit-based interface to process uploaded farm images and videos.
The application is designed to:
- Identify rice varieties from uploaded images.
- Analyse uploaded farm videos using sampled frames.
- Generate prediction summaries and downloadable CSV reports.
- Display model confidence scores for predicted rice varieties.
- Python
- Streamlit
- Ultralytics YOLO
- OpenCV
- Pandas
- Pillow
The project uses a trained YOLO classification model covering 33 rice variety classes.
CropAI is under active development. This repository will document the project's implementation, model integration, evaluation and improvements progressively.
Model confidence scores are predictions, not guarantees of real-world accuracy. Performance can vary with image quality, lighting, camera angle and field conditions.