Field-Ready Robotics: Sensor- Based Crop Monitoring and Precision Farming
Primary Investigator (PI) Name
Muhammad Hassan Tanveer
Department
SPCEET – Robotics and Mechatronics Engineering
Abstract
Monitoring the health of crops is a critical part of ensuring consistent and high-quality output for farmers. A powerful method for evaluating the health of crops is Multispectral imaging, which uses a Red-Green-Near-Infrared camera to extract a variety of visual health indicators known as vegetative indices. Although this approach is effective, it is not available to all farmers as the cost of field ready multispectral cameras can often be out of budget. In our research, we aim to tackle this problem by extracting RGN vegetative indices from Red Green images. We implement an Image Generation Model based on Artificial Neural Networks, using captured Red-Green-Near-Infrared images and their subsequent Red-Green images to train the model. The specific use case evaluated is predicting the Normalized Difference Vegetative Index (NDVI) result of a Red-Green image, as the NDVI is the most common index utilized for visual health monitoring. Ultimately, this generative approach would allow for the use of a consumer grade Red-Green-Blue camera to collect pertinent crop health information. This results in savings for the farmers while still being able to monitor the health of their crops effectively.
Disciplines
Robotics
Field-Ready Robotics: Sensor- Based Crop Monitoring and Precision Farming
Monitoring the health of crops is a critical part of ensuring consistent and high-quality output for farmers. A powerful method for evaluating the health of crops is Multispectral imaging, which uses a Red-Green-Near-Infrared camera to extract a variety of visual health indicators known as vegetative indices. Although this approach is effective, it is not available to all farmers as the cost of field ready multispectral cameras can often be out of budget. In our research, we aim to tackle this problem by extracting RGN vegetative indices from Red Green images. We implement an Image Generation Model based on Artificial Neural Networks, using captured Red-Green-Near-Infrared images and their subsequent Red-Green images to train the model. The specific use case evaluated is predicting the Normalized Difference Vegetative Index (NDVI) result of a Red-Green image, as the NDVI is the most common index utilized for visual health monitoring. Ultimately, this generative approach would allow for the use of a consumer grade Red-Green-Blue camera to collect pertinent crop health information. This results in savings for the farmers while still being able to monitor the health of their crops effectively.