Artificial Intelligence based Efficient Phenotyping for Agronomics

The evaluation of soil and plants has been a costly, labor-intensive, and time-consuming process traditionally. With the advancements in technology, the phenotypic variations in plants and their corresponding information can be extracted efficiently using IoT, WSN, and computer vision and processed with machine learning, artificial intelligence, and such cutting-edge techniques. This can help in accelerating the developments in vegetation, improving the genomic selection, and enable high-throughput plant phenotyping. In this work, the phenotypic characteristics of the crop and the soil conditions are assessed using deep learning convolutional neural networks based on the data obtained by the IoT sensors and multispectral imaging from small Unmanned Aerial Vehicles (UAVs) across the agricultural area. The plant quality and leaf diseases are identified. 99.9% precision and recall are obtained with an overall accuracy of 95%. This technique is quick, cost-effective, and reliable.