Vision-based lettuce phenotype modeling using computational intelligence

The loose-leaf lettuce (Lactuca sativa var. Altima and var. Rania) is an important food security crop for developing world. Erratic extreme environmental conditions and the potency of available water nutrients are the major constraints to yield L. sativa in the tropical region. Controlled environment agriculture may be the most pragmatic solution to mitigate the yield deficits introduced by the nonlinear uncontrolled environment. However, sustainability in plant factories is the core issue associated with it. In this study, a vision-based aquaponic lettuce phenotype (VIPHLET) model consisting of crop phenology, growth, and quality modules was developed using computational intelligence to non-destructively quantify and extract lettuce phenotype signatures and set it as input to the adaptive fertigation system and fabrication of an organic photovoltaic cell. The following work utilizes in vivo and controlled nutrient film technique aquaponic trials to optimize the abiotic pre-harvest growth factors, characterize architectural and anatomical root, stem, and leaf phenes in Lactuca species, and quantify their interaction to its environment. The results of this work suggest that: hybrid aquaphotomics and genetic programming is effective in determining nitrate, phosphate, and potassium concentration in pond water using physico-limnological sensors substitution; evolutionary strategy-optimized anthocyanin and chlorophyll a concentration with ethanol extract solvent provides a 24.093% photon utilization efficiency in lettuce leaf extract-sensitized photovoltaic cell; thermo-gas dynamics revealed that uneven broadening of leaf canopy allows high oxygen production, high full moisture content, and low equivalent water thickness; artificial white light induces chlorophyll a, beta-carotene, anthocyanin, lutein, and vitamin C concentrations while red-blue light spectrum promotes chlorophyll b; multiverse and firefly optimization algorithms determined the ideal abiotic pre-harvest growth factors of 880.744 ppm of carbon dioxide, 543.147 μmol m-2 s-1 of white light spectrum, 22.238 °C air temperature, 69.742% humidity, 204.10 mg L-1 nitrate, 238.15 mg L-1 phosphate, and 158.08 mg L-1 potassium concentrations; and finally, the use of VIPHLET model allied with the Mamdani fuzzy logic-based nutrient tank solenoid valve controller for adaptive fertigation system resulted in higher nutrient use efficiency that is approaching 100% and lower chemical waste emission of 14.108 mg of NPK fertilizer L-1 of water than the conducted manual fertigation over a 6-week cultivation period. Overall, this study offers substantial and proven technologies for sustainable precision agriculture in an adaptive fertigation system and clean food production system in terms of means of energy resource accumulation