Abstract
Phosphorus is an important macronutrient for plant development, but its bioavailability in soil is often limited. Phosphate-solubilizing microorganisms play a vital role in phosphorus biogeochemistry, offering a sustainable alternative to chemical fertilizers, which pose environmental risks. Manual measurements for quantifying phosphate solubilization capacity are laborious, subjective, and time-consuming, so there is a need to develop more efficient and objective approaches. This study aimed to develop and validate a machine vision system called IGLOO to automate and optimize the determination of relative phosphate solubilization efficiency in phosphate-solubilizing bacteria. IGLOO was developed using YOLOv8 in conjunction with creating and labeling a dataset of images of bacterial colonies grown in vitro with the bacterial strains Enterobacter R11 and FCRK4. The model was trained with a different number of epochs. IGLOO's performance was evaluated by comparing its segmentation accuracy with accepted metrics in the domain and by contrasting its solubilization efficiency estimates with experts' manual measurements. The model achieved greater than 90% accuracy for colony and halo detection, with a relative error of less than 6% compared to manual measurements, demonstrating its reliability by minimizing observer variability. Finally, IGLOO represents a significant advance in the quantitative evaluation of phosphate solubilization of microorganisms because it reduces analysis time and provides objective and reproducible results for agricultural studies.