The IBEM dataset: A large printed scientific image dataset for indexing and searching mathematical expressions

•The new IBEM dataset has 160000+ mathematical expressions extracted from 600 documents containing 8200 page-images.•The IBEM dataset includes ground truth consisting of formula page level location and associated LaTex markup transcription.•The IBEM dataset use cases are mathematical expression detection, extraction, recognition, and retrieval.•The paper provides benchmark results on symbol classification and mathematical expression recognition for comparison purpose.