Event classification is a common task in gamma-ray astrophysics. It can be treated with rapidly-advancing machine learning algorithms, which have the potential to outperform traditional analysis methods. However, a major challenge for machine learning models is extracting reliably labelled training examples from real data. Citizen science offers a promising approach to tackle this challenge.
We present "Muon Hunter", a citizen science project hosted on the Zooniverse platform, where VERITAS data are classified multiple times by individual users in order to select and parameterize muon events, a product from cosmic ray induced showers. We use this dataset to train and validate a convolutional neural-network model to identify muon events for use in monitoring and calibration. The results of this work and our experience of using the Zooniverse are presented.