Systems and methods include techniques for using smart polymers. Units of smart polymers with hydrogen sulfide (H2S) sensitivity are inserted by a monitoring system into drilling fluid pumped into a well. The smart polymers are configured to be triggered by increasing H2S concentrations. An insertion timestamp associated with each unit is stored. Each insertion timestamp indicates a time that each unit was inserted. Continuous images and observed characteristics of returning mud exiting through an annulus of the well and containing the units of smart polymer are captured by a camera positioned at a sensing location and linked to the monitoring system. An estimate of H2S levels at a drill bit of the drilling operation is determined using continuous images, observed characteristics, and insertion timestamps, and based at least in part on executing image processing algorithms, machine-learning models, and deep-learning models. Changes to drilling parameters are suggested.