Overview / Usage

I worked at a student lab where we used multispectral cameras sponsored by our college and software by © Parrot SA in our initial build and we will be doing it with regular low cost drone cams in the future builds to reduce the cost even further. This would be done using advanced Deep Learning techniques and the mapping and navigation would be automated using a mixture of techniques like Optical Flow, SLAM, Reinforcement Learning etc. I have also used Intel's RealSense SDK and their cameras to aid our research. The multispectral images given by the camera were index maps, like NDVI, and uses RGB images to generate high resolution orthomosaics which can be easily used to monitor the health. The image attached below shows a sample multispectral image. For reducing the costs, a regular camera with sensor modification would also work but for best results an industry grade multispec camera is recommended. The processing done on the image can be done on a cloud based solution which could run through a python script or could as well be done natively by running scripts to calculate NDVI and other indices. I will share useful resources in the links below, you just need to adjust and optimize the code according to the dataset you collect using the camera.