AI-Fission Track-Training Image Dataset

We introduce the AI-Fission-Track Testing Dataset (Boone et al., in review), which was developed to benchmark the HALtracks convolutional neural networks against both traditional and other AI-based automated fission-track counting methods. The dataset contains reflected-light images and transmitted-light z-stacks of spontaneous fission tracks in diverse apatite grains that span a broad spectrum of ages, track densities, confined track length distributions, and Dpar values, together with induced tracks recorded in uncoated mica. Additional information is provided in Boone et al. (in review).