PhD Candidate at University of Bordeaux in France.
Axel Ransinangue is a Ph.D. candidate at Bordeaux University, conducting research at the intersection of artificial intelligence and geosciences. Specializing in deep learning for carbonate reservoir characterization, his work integrates advanced image processing, computer vision, and geological interpretation. Axel’s expertise spans Python, MATLAB, TensorFlow, PyTorch, and geospatial tools such as QGIS and ArcGIS, enabling him to develop innovative solutions for analyzing thin section images, petrophysical properties, and hyperspectral datasets. Collaborating closely with TotalEnergies, he has designed semi-supervised classification systems, synthetic data generation pipelines, and multiscale segmentation techniques that bridge synthetic and real-world geological imagery. His contributions extend beyond research, actively engaging in scientific communication through conferences and leading discussions in the computer vision community. Driven by a passion for data-driven science, Axel’s work demonstrates both academic rigor and industrial relevance, making him a promising leader in AI-driven geoscience innovation.
Professional Profile
Google Scholar
Education
Axel holds a Bachelor’s degree in Earth Sciences and Environment with honors from Pau University, where he developed strong foundations in porous media analysis and image processing. He pursued a Master’s degree in Engineering from Bordeaux INP – ENSEGID, graduating with honors, and participated in an international exchange at Kyushu University, Japan, expanding his technical and cultural perspectives. Currently, Axel is a Ph.D. candidate in Artificial Intelligence, Sciences, and Environment at Bordeaux University, working in collaboration with TotalEnergies. His doctoral research integrates AI methodologies with carbonate petrography to enhance reservoir characterization. Under the supervision of experts in geology and computer science, he specializes in representation learning, domain adaptation, and synthetic data conditioning for geological imagery. This interdisciplinary education has equipped him with a unique blend of computational, analytical, and field-specific skills, positioning him at the forefront of AI applications in earth sciences.
Experience
Axel’s professional experience blends academic research with industrial applications. At TotalEnergies, as a Geologist Intern, he analyzed carbonate thin sections, interpreting petrographic features for reservoir evaluation. Later, as a Data Scientist at AGEOS, he applied hyperspectral imaging to mineralogical quantification, developing regression models and calibrating point cloud acquisitions. His current role as a Ph.D. researcher involves designing deep learning systems for carbonate reservoir characterization, focusing on semi-supervised classification, conditional synthetic dataset generation, and multiscale image segmentation. He has also explored model explainability, ensuring AI decisions are interpretable for geological experts. Additionally, Axel has worked on integrating bi-modal classification models combining imagery with petrophysical data, as well as anomaly detection frameworks. His cross-domain expertise enables the translation of AI methodologies into practical tools for geoscience, creating value both in research and industrial operations.
Research Focus
Axel’s research lies at the convergence of computer vision, deep learning, and carbonate petrography. His primary objective is to enhance geological image analysis through advanced AI-driven methodologies. Key areas include representation learning with invariance to interpretation biases, synthetic dataset generation conditioned on geological parameters, and domain adaptation to bridge synthetic and real-world imagery. He specializes in texture synthesis, pixel harmonization, and object packing strategies for creating high-quality training data when labeled datasets are scarce. His work also involves developing heuristics-based regularization techniques for improved segmentation accuracy and integrating statistical analysis to link image descriptors with petrophysical properties. By leveraging semi-supervised and multi-modal approaches, Axel aims to create robust and generalizable AI models that address challenges in reservoir characterization. This research not only advances geological sciences but also contributes to broader AI applications in image-based data interpretation across environmental and industrial domains.
Publication Top Notes
Title: SynSection: Sedimentology-driven data generation for deep learning applications in carbonate petrography
Authors: A. Ransinangue, R. Labourdette, E. Houzay, S. Guillon, R. Bourillot, et al.
Journal: Marine and Petroleum Geology, Article ID 107490.
Summary: The study presents SynSection, a framework for generating synthetic carbonate thin section images based on sedimentological parameters. Combining texture synthesis, object packing, and pixel harmonization, it produces realistic datasets to train deep learning models when labeled geological data is scarce. Demonstrated improvements in image classification and segmentation highlight its potential for reservoir characterization and data-driven petrography.
Conclusion
Axel Ransinangue presents a compelling case for recognition as a Best Academic Researcher. The work combines cutting-edge AI methodologies with domain-specific geological expertise, producing research that is both academically valuable and industrially applicable. With ongoing expansion of publication output and interdisciplinary collaborations, the candidate has strong potential to emerge as a leading figure in AI-driven geoscience research. Their contributions already reflect a blend of innovation, rigor, and practical relevance that aligns well with the award’s intent.
