Facilitating the Identification of the Nannofossil Species in Cretaceous of Abu Dhabi Using Artificial Intelligence

Abstract Reliable chrono-stratigraphic correlation is crucial in the reservoir evaluation. Lithostratigraphic correlation is often inconsistent with depositional age due to high heterogeneity in lithology. Geological age determination based on nannofossils in Middle East gives a better control for the stratigraphic correlation, however it requires skilled experts and time for appraisal. This contribution explores if today's emerging AI technology provides a solution, by facilitating the identification of the nannofossil species on smear slide images. The result of our exercises shows that a deep-learning-based object detector reasonably and quickly identifies nannofossil species on micrographs and can deliver information useful for the geological age assessment. We find that separating detection and classification procedures improves the performance of the object detector. Further data collection and model tuning of the object detector are needed for the improvement of the object detection accuracy, especially for nannofossil species rarely observable.