The popularisation of neural networks has seen incredible advances in pattern recognition, driven by the supervised learning of human annotations. However, this approach is unsustainable in relation to the dramatically increasing size of real-world datasets. This has led to a resurgence in self-supervised learning, a paradigm whereby the model generates its own supervisory signal from the data. Here we propose a hybrid quantum-classical neural network architecture for contrastive self-supervised learning and test its effectiveness in proof-of-principle experiments. Interestingly, we observe a numerical advantage for the learning of visual representations using small-scale quantum neural networks over equivalently structured classical networks, even when the quantum circuits are sampled with only 100 shots. Furthermore, we apply our best quantum model to classify unseen images on the ibmq_paris quantum computer and find that current noisy devices can already achieve equal accuracy to the equivalent classical model on downstream tasks.