Identification of Plant Health Using Machine Learning and Image Techniques

Farming has become increasingly risky with uncertain natural growth processes of crops, weather, disease, pests, other site-specific factors, and lack of support services. The production risk is apparent. Currently, farmers are using manual, labor intensive methods to assess in-situ variations based on their experience and knowledge. There are numerous precision agronomy tools available in the market for balancing field variations and healthy growth. Mostly they are electronic sensors that attach to machinery to take remotely sensed measurements. This research is to develop an application to measure crop health using low-cost remotely sensed images. The project used a camera that is sensitive to visible and near-infrared regions of the electromagnetic. The camera provided three-band images (green, red and near-infrared). The images collected from the field were pre-processed and separated the plants and the background. The images were fed into the model for training for identifying the health of the plant.The results were based on the training data included in the model earlier. Later, the model was developed as a mobile application that can be used by farmers