We present results on the reconstruction of electromagnetic (EM) activity from photons produced in charged current ν _μ interactions with final state π ⁰s. We employ a fully-automated reconstruction chain capable of identifying EM showers of (100) MeV energy, relying on a combination of traditional reconstruction techniques together with novel machine-learning approaches. These studies demonstrate good energy resolution, and good agreement between data and simulation, relying on the reconstructed invariant π ⁰ mass and other photon distributions for validation. The reconstruction techniques developed are applied to a selection of ν _μ + Ar → μ + π ⁰ + X candidate events to demonstrate the potential for calorimetric separation of photons from electrons and reconstruction of π ⁰ kinematics.