Graphene has been shown to be a prospective platform for active photonic devices with exotic optical properties. Employed in these devices, the graphene photodoping mechanism would allow for the remote spatiotemporal doping control by means of illumination, not restricted by the physical gate electrodes. This paper reports on the efficient graphene photodoping in graphene-CH3NH3PbI3 perovskite heterostructure recently spotlighted for photodetector applications. To maximize the photoresponse, the heterostructure is optimized by systematically introducing additional layers of the self-assembled monolayer of octadecyltrichlorosilane molecules, MoO3, and poly(3,4-ethylenedioxythiophene) polystyrene sulfonate for improving interfacial charge transfer properties. The photodoping amount estimated from the photocurrent measurements is 2.7 x 10(12) cm(-2) at zero gate biasmore than twice higher than previously observed in graphene-perovskite heterostructures. Without external gate bias, the estimated graphene Fermi energy varies from 0.37 to 0.43 eV due to photodoping, which is sufficient to operate graphene-based active photonic devices at mid-infrared frequencies, and similar to that typically achievable with the conventional electrostatic gating. Furthermore, this work highlights the missing yet important aspect of the characterization method for the phototransistors with a 2D channel, resulting from the mismatch between the units of the current and power densities.