This image dataset supports the findings in our manuscript entitled Morphological deconvolution of beta-lactam polyspecificity in E. coli. The images display fluorescently-labeled E. coli strains subjected to compound treatment at multiple concentrations. Cells were imaged in 96-well plates at 100X magnification. Our novel method for preparing E. coli cells for imaging enabled us to use commercially-available plate-based, automated HCI microscopes. We used the images to build our deep neural network classifier to identify bacterial morphologies as well as our Hidden Markov models for target deconvolution.