We address the task of detecting bacteria and estimating bacterial load in the human distal lung with fibered confocal fluorescence microscopy (FCFM) and a targeted smartprobe. Bacteria appear as bright dots in the image when exposed to a smartprobe, but they are often difficult to detect due to the presence of background autofluorescence inherent to human lungs. In this study, we create a database of annotated image frames where a clinician has labelled bacteria, and use this database for supervised learning to build a suitable bacterial load estimation software.