Οδικός Χάρτης Τεχνητής Νοημοσύνης για Επιχειρήσεις Agriculture στην Split

Agriculture in the Split-Dalmatia County is characterized by fragmented, non-contiguous plots and rugged 'Karst' topography. Standard satellite-based NDVI models often fail due to the high density of limestone and rocky outcrops. Our approach involves deploying Edge AI on specialized drones using multispectral sensors specifically tuned to the spectral signatures of Plavac Mali grapes and Oblica olives. By utilizing localized training sets that account for the high-reflectance karst background, we enable precise leaf-area index (LAI) calculations and early detection of Peacock Spot (Spilocaea oleaginea) which are often missed by generic agricultural software.

With increasing summer heatwaves in the Adriatic, water management is a critical pain point for Split's agricultural sector. We implement Reinforcement Learning (RL) algorithms coupled with LoRaWAN-enabled soil moisture sensors. Unlike traditional scheduled irrigation, these agents autonomously adjust water delivery based on evapotranspiration rates and forecasted 'Bura' wind patterns—which accelerate soil drying. Furthermore, in coastal areas near the Kaštela Bay, the system monitors for saltwater intrusion, triggering automated mitigation strategies to protect the root zones of high-value aromatic herbs from salinity stress.