This report presents the work of a 6-month internship part of the MSc in Modeling for Neuronal and Cognitive systems. This work explores possible approaches in the use of simulated Retinal Waves (RWs) as pre-training for machine learning (ML) computer vision models in Optical Flow Estimation (OFE). Retinal Waves are one of the early processes of visual system development in mammals, structuring the retinal connectivity and thus preparing for vision, including motion detection. We select a recent, non-Transformer ML architecture, RAFT, and adopt a Transfer Learning strategy to leverage RWs in enhancing OFE performance through a related task. Additionally, we explore an alternative approach that estimates an approximated optical flow of RWs, allowing for its direct application within OFE. The idea of using these biological stimuli to generate more accessible training data to improve the generalization capabilities of OFE models shows a limited effectiveness with both approaches.