Page 213 - AI for Good-Innovate for Impact Final Report 2024
P. 213

AI for Good-Innovate for Impact



               Use case – 50: Using Satellite Imagery and AI for Crop Monitoring

               Services                                                                                             50-AIPARAGRO








               Country: Zimbabwe

               Organization: abm^parassa IT/Digital Consulting

               Contact person: Abel Mautsa abelmautsa@ gmail .com, +263713852018

                                Benhilda Mubaiwa benhildamubaiwa@ gmail .com ,+ 263773731535


               50�1� Use case Summary Table


                Domain              Agriculture
                The problem to be   Crop monitoring using satellite image, complemented with drone
                addressed           images. Small scale farmers, and medium to large scale farmers
                                    Production is labour intensive. Time investment, human error, timely
                                    detection of crop related issues.
                Key aspects of the   Using Satellite, UAV and scouts imagery to monitor crop for irrigation,
                solution            plant nutrition, pest/weeds mangement and yield prediction.
                                    Farmer connectivity is via mobile data via 3G, 4G, or VSAT links or fiber.
                                    But image transmission is via telone.

                Technology          Satellite, UAV and Scouts imagery, Field boundary detection, NDRE,
                keywords            MSAVI, NDWI, NDMI, NDVI, Neural Network model.

                Data availability   Not available.
                                    Current solution is based on the eos Data Analytics [1] (eos data analyt-
                                    ics - satellite data analytics provider based in Ukraine) and telone
                                    partnership.
                                    Need to acquire satellite images, and drones images from farmers.
                Metadata (type of   Images (satellite or from drones)
                data)               Sensor data.

                Model Training and   The current commercial solution with eosda and telone uses different
                fine-tuning         type of models.

                Testbeds or pilot   •  https:// eos .com/ blog/ eosda -enters -into -a -strategic -partnership
                deployments            -with -telone/
                                    •  https:// eos .com/ products/ high -resolution -images/
                                    •  https:// eos .com/ products/ crop -monitoring/














                                                                                                    197
   208   209   210   211   212   213   214   215   216   217   218