High-density single-particle tracking based on optical flow learning

This is the official dataset of high-density particle tracking using a video flow-informed neural network (VFINN) that estimates dense optical flow without requiring ground-truth data. VFINN fine-tunes a large visual neural network with a flow-informed loss function, achieving sub-pixel average endpoint error. This dataset provides formatted single-particle tracking .tif image pairs with localization points, ready for direct use in training.