All the time, common pathogens in urine have been threatening human health even lives. Though Chromogenic plates is able to detect urinary bacteria, another technique and material need to be introduced for coloration. However, this kind of technique and material is not always specific. Therefore, how to rapidly detect and classify these bacteria is a tricky challenge waiting to be overcome. For solving this problem, we use hyperspectral imaging (HSI) and random forest (RF) to classify four strains of common pathogens in the urinary, which include E.faeca, K.pneum, S.aureu, and S.pneum. HSI is able to obtain spatial and hyperspectral information of the object. RF boosts classifying accuracy by combining many decision trees. Utilizing HSI and RF, we find that these bacteria can be separated well at the cellular level. © 2021 IEEE.