Currently, most agricultural imaging tools rely on satellite images that can utilize a powerful zoom function to detect signs of pests, dryness, and disease. Further, these technologies are catered towards larger commercial farmers that would work for a larger cooperative.
This technology provides a non-contact, non-invasive, and quantitative solution for monitoring plant health by leveraging optical biospeckle imaging combined with machine learning analysis. The system operates by illuminating plant material with coherent light, such as from a laser, and capturing the resulting dynamic biospeckle patterns using an image sensor. These patterns, which reflect underlying biological activity within the plant, are preprocessed to extract bioactivity metrics through temporal speckle contrast algorithms. The processed data is then analyzed by a trained machine learning model that quantitatively assesses plant health, identifying issues such as disease, dehydration, or pest presence. The system’s modular architecture includes a light source, optics, image sensor, processor, memory, and communication interfaces, allowing it to be deployed flexibly on drones, tractors, handheld devices, or smartphones for both large-scale and targeted agricultural monitoring.
Preliminary results for a leaf of Capsicum annuum