This repository contains code and models for a comprehensive system that leverages images for crop-related tasks. The system is divided into three main modules:
-
Crop Identification: Utilizing images captured by high resolution cameras, this model can identify various crops such as rice, maize, wheat, sugarcane, and paddy. The accuracy of this model is an impressive 91.9%.
-
Fire Detection: The second module focuses on fire detection using image processing techniques. The model can analyze images to identify and locate fires in agricultural fields. The accuracy of this model is 95%.
-
Crop Yield Prediction: The third module predicts crop yield based on the collected data. It incorporates information from the crop identification module and provides accurate predictions with a high accuracy of 96.5%. Additionally, the system offers valuable suggestions for improving production based on the provided data.
-
Aerial Seeding: Another feature of this project is of aerial seeding in which there is a system using which seeds can be dropped in the farm using the aerial seeding system. The system will be provided 30 seconds to position itself and then the seed box will continue to open and close at intervals of 5 seconds for 20 to 30 minutes. The code for the arduino file has been provided in the repository
The crop identification model was trained on a diverse dataset containing images of rice, maize, wheat, sugarcane, and paddy. You can find the dataset here.
The fire detection model relies on a dataset specifically curated for identifying fires in agricultural fields. Access the dataset here.
The crop yield prediction model was trained on a dataset encompassing various factors influencing crop yield. You can access the dataset in the repository itself
- Crop Identification Model Accuracy: 91.9%
- Fire Detection Model Accuracy: 95%
- Crop Yield Prediction Model Accuracy: 96.5%
- Clone the repository:
git clone https://github.com/your-username/crop-classification-and-yield-prediction.git