Turfgrass divot maintenance is a vital component of sports grounds and golf courses. Preci- sion Turfgrass Management (PTM) uses a variety of smart technologies to manage turfgrass in a sustainable and efficient manner. Deep Learning is an effective way of developing data inspection but requires a large, annotated dataset. This work describes the development of a photo realistic dataset generated using Blender for a turfgrass environment to train deep learning networks to iden- tify anomalies on turfgrass surfaces. Additionally, two colour transfer techniques are analysed, to ensure the synthetic data can closely resemble the real domain data.