Urban street trees depend on storm water infiltration, which is strongly influenced by the type of pavement surrounding them. This dataset supports the development of scalable, image-based methods for pavement mapping at the individual tree level. It contains geo-referenced close-range image sequences collected around street trees in Sacramento and Davis, California (USA). The images are intended for the generation of orthoprojections and subsequent classification of pavement materials above the root zone of urban street trees. Data collection was performed by walking around each tree with a camera setup, capturing overlapping images of the pavement and geo-referenced markers in each scene that can be processed into orthoprojections using photogrammetry.