During machining, material shears at strains and strain rates that are of several orders of magnitude higher than what the same material would have experienced during typical tensile experiments. Since deformation mechanisms are fundamental to how the material will behave during its manufacture and use in an application, characterizing deformation behavior is essential to develop flow stress and damage models that could be used for predictive purposes. This is only possible by visualizing the machining process using vision-based in situ monitoring. This paper discusses a framework in which we use a digital image correlation scheme to estimate flow behavior from video of a cutting process. We illustrate the method on video of orthogonal cutting of a hard plastic material with different cutting conditions. We discuss the noise floor and sensitivity of the estimates to image processing and acquisition parameters. We benchmark estimations from video with classical theoretical models. Direct estimates of shear angles are found to agree with model predictions. Estimates for strain rates are observed to increase with cutting velocity, whereas indirect estimates for friction are observed to reduce with speed. Since estimates follow expected trends, our framework provides a blueprint for vision-based in situ characterization of material deformation during machining processes.