Operational usability of {Deep} {Learning} methods for automated microfossil detection on thin sections: ongoing assessment and methodological insights

@inproceedings{bertaud_operational_RM2023,
 abstract = {In the last decade, Deep Learning methods have been explored to automate and accelerate tedious, timeconsuming, or hard-to-delegate tasks in multiple scientific and industrial domains. Geoscience is no exception and Deep Learning approaches have notably been tested with promising results for various image analysis processes, for instance with seismic profiles, core photographs or thin section scans. However, these techniques are not yet fully adopted by operational geoscientists in the subsurface industries, and their usability in routine activities still needs to be demonstrated. In particular, classical Deep Learning recipes often need to be questioned and adjusted to maximize their relevance when dealing with operational geoscience use cases.},
 author = {Bertaud, Lucie and Bouziat, Antoine and Ammar, Aziz Ben and Hamon, Youri and Ferraille, Mathieu},
 booktitle = {2023 {RING} meeting},
 language = {en},
 pages = {13},
 publisher = {ASGA},
 title = {Operational usability of {Deep} {Learning} methods for automated microfossil detection on thin sections: ongoing assessment and methodological insights},
 year = {2023}
}