Dataset for associated publication. Abstract: Point defect identification in two-dimensional materials enables an understanding of the local environment within a given system, where scanning probe microscopy that takes advantage of hyperspectral tunneling bias spectroscopy acquisition can both map and identify the atomic and electronic landscape. Here, dense spectroscopic volume is collected autonomously via Gaussian process regression, where convolutional neural networks are used in tandem for defect identification. Acquired data enable image segmentation across defect modes, and a workflow is provided for both machine-driven decision making during experimentation and the capability for user customization. Where scanning tunneling microscopy and spectroscopy can be nontrivial and time intensive, we provide a means towards autonomous experimentation for the benefit of both enhanced reproducibility and user-accessibility. Monolayer semiconductor is explored on tungsten disulfide sulfur vacancies to provide two-dimensional hyperspectral insight into available sulfur-substitution sites within a transition metal dichalcogenide (TMD), which is combined with spectral confirmation on the Au{111} herringbone reconstruction for both tip state verification and local fingerprinting. Overall, chalcogen vacancies, pristine TMD, Au face-centered cubic, and Au hexagonal close packed regions are examined and detected by machine learning methods.