Farm robots currently struggle in scattered fields due to a lack of precise map data. To solve this, we created a system using satellite imagery and AI to automatically generate accurate field boundaries. Our tests showed 93 % accuracy across different regions. This technology serves as "digital eyes" for machinery, replacing slow manual inputs with automated data. It enables robots to navigate and harvest continuously in complex smallholder farms, unlocking the full potential of smart agriculture.