Multispectral images for discrimination of sources and doses of fertilizer in coffee plants

Remote monitoring of the management coffee crops is necessary as demand in decision-making, where aim to rise production based on sustainable a constant growth. In this work, it was evaluated potential images obtained by low-cost sensors discrimination sources and doses mineral organomineral fertilizers coffee. The experimental design randomized blocks, with five blocks six treatments, follows: (T1) - 100% treatment; (T2) 70% (T3) 50% (T4) fertilization; (T5) standard treatment farm (T6) fertilization. After management, we used Mapir 3 Survey3W camera coupled an ARP drone – Phantom4 take experiment over 12-month vegetative period. Combined image taking, collected agronomic parameters growth productivity for two concluded that different fertilization did not significantly affect analyzed parameters. Based supervised classification multispectral images, possible discriminate treatments higher degree accuracy (86.66% accuracy) than when analyzing