Léonard Martinez

Léonard Martinez

Auto-generated profile, based on publicly available information

Léonard Martinez

Université Paris‐Saclay, CNRS, GEOPS Orsay France

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 March 2025 in Planetary and Space Science
Verified
Journal Article
Closed access
LML. Martinez
L. Martinez
FAF. Andrieu
F. Andrieu
FSF. Schmidt
F. Schmidt
HTH. Talbot
H. Talbot
MBM.S. Bentley
M.S. Bentley

3 July 2024
Verified
Conference Paper
Open access
LMLeo Martinez
Leo Martinez
FSFrédéric Schmidt
Frédéric Schmidt
FAFrançois Andrieu
François Andrieu
MBMark Bentley
Mark Bentley
HTHugues Talbot
Hugues Talbot
Introduction: Understanding the distribution and characteristics of impact craters on planetary surfaces is essential if we want to understand the geological processes such as tectonic deformation, volcanism1, erosion, transport, and impact cratering itself2 which constantly rebuilt planetary surfaces. By analysing the distribution and the density of crater using the "crater counting" approach, it is possible to estimate the age of the planetary surface3 at regional scale. Historically, crater detection has been made manually. For Mars and the Moon, the existing handmade databases4,5,6 brings an inestimable value to the community. Nevertheless, even with corrective analysis7, manual databases are subject to human limitations. Indeed, studies have shown that human attention to repetitive tasks, such as crater counting, rapidly decreases after 30 minutes8and then errors started to occur. Several machine learning and AI-based approaches have been proposed to automatically detect craters on planetary surface images9,10,11. Data: To train our machine learning algorithm, we basically need two kind of data. We need images for the detection and a global crater database which encompass the ground truth. In this work we took images from the context camera on board of the Mars Reconnaissance Orbiter (CTX) and preprocess by the Bruce Murray Laboratory of Caltech12. For the crater database we took A. Lagain’s global martian database which encompass more than 376000 craters of a size bigger than 1 km in diameter7. Method: In this work, we presents a novel approach using the Faster Region-based Convolutional Neural Network (Faster R-CNN) for automatic crater detection13. As shown in Fig. 1, Faster R-CNN employs three stages: CNN backbone extracts features, RPN generates ROIs with anchor boxes predicting object presence and adjustments, and Detector refines object classification and bounding box coordinates, ensuring precise object detection and localization in images.Fig. 1 : Faster R-CNN architecture as describe by S.Ren & al, 201613 with our image configurationThe proposed method involves a preprocessing step in which we cut the images to a size of 224x224 pixels, we reproject the images in order to be sure that crater will always have a circular shape at ever latitude and we split the crater database to have a ground truth label for each image. Then, we train our model with 82874 images and we test our detector on 4828 images.Results: Extensive experiments on high-resolution planetary imagery demonstrate excellent performances with a mean average precision mAP50 > 0.82 with an intersection over union criterion IoU ≥ 0.5, independent of crater scale (Fig.2 & 3).Fig. 2 : Precision as a function of recall for six different IoU thresholds. Fig. 3 : Precision vs Recall curves for an IoU = 0.5 and for different bounding box sizes of the test dataset.The Fig. 4 show an example on an equatorial Mars quadrangle. Fig. 4 : Inference on an equatorial region of Mars. The lower right corner of this 4°×4°quadrangle is located at 44°W, 0°N. The yellow boxes represent the ground truth infor-mation and the blue boxes the predictions ones. Pleasenote that crater smaller than 10 pixels in diameter are ignored. The results also highlight the versatility and potential of our robust model for automating the analysis of craters across different celestial bodies. The automatic crater detection tool holds great promise for future scientific research of space exploration missions. References: [1] : M. H. Carr, Volcanism on mars 78 (20) 4049–4062. doi:10.1029/ jb078i020p04049.[2] : Hartmann, William K., and Gerhard Neukum. "Cratering chronology and the evolution of Mars." Chronology and Evolution of Mars: Proceedings of an ISSI Workshop, 10–14 April 2000, Bern, Switzerland. Springer Netherlands, 2001.[3] : G. Neukum, B. A. Ivanov, W. K. Hartmann, Cratering records in the in- ner solar system in relation to the lunar reference system, in: Chronology and Evolution of Mars: Proceedings of an ISSI Workshop, 10–14 April 2000, Bern, Switzerland, Springer, 2001, pp. 55–86.[4] : Head III, James W., et al. "Global distribution of large lunar craters: Implications for resurfacing and impactor populations." science 329.5998 (2010): 1504-1507.[5] : S. J. Robbins, A new global database of lunar impact craters¿ 1–2 km: 1. crater locations and sizes, comparisons with published databases, and global analysis, Journal of Geophysical Research: Planets 124 (4) (2019) 871–892[6] : S. J. Robbins, B. M. Hynek, A new global database of mars impact craters ≥ 1 km: 1. database creation, properties, and parameters, Jour- nal of Geophysical Research: Planets 117 (E5) (2012).[7] : A. Lagain, S. Bouley, D. Baratoux, C. Marmo, F. Costard, O. De- laa, A. P. Rossi, M. Minin, G. Benedix, M. Ciocco, et al., Mars crater database: A participative project for the classification of the morpho- logical characteristics of large martian craters, geoscienceworld (2021).[8] : J. E. See, S. R. Howe, J. S. Warm, W. N. Dember, Meta-analysis of the sensitivity decrement in vigilance., Psychological bulletin 117 (2) (1995) 230.[9] : Salamuniccar, Goran, and Sven Loncaric. "Method for crater detection from martian digital topography data using gradient value/orientation, morphometry, vote analysis, slip tuning, and calibration." IEEE transactions on Geoscience and Remote Sensing 48.5 (2010): 2317-2329.[10] : G. Benedix, A. Lagain, K. Chai, S. Meka, S. Anderson, C. Norman, P. Bland, J. Paxman, M. Towner, T. Tan, Deriving surface ages on mars using automated crater counting, Earth and Space Science 7 (3) (2020) e2019EA001005.[11] : R. La Grassa, G. Cremonese, I. Gallo, C. Re, E. Martellato, YOLOLens: A deep learning model based on super-resolution to enhance the crater detection of the planetary surfaces, Remote Sensing 15 (5) (2023) 1171. doi:10.3390/rs15051171.[12] : Dickson,...

9 July 2025
Verified
Conference Paper
Open access
LMLeo Martinez
Leo Martinez
FAFrançois Andrieu
François Andrieu
FSFrédéric Schmidt
Frédéric Schmidt
MBMark S. Bentley
Mark S. Bentley
Introduction Impact craters are essential markers for reconstructing the geological history of planetary surfaces [1]. On Mars, where no absolute radiometric dating has yet been conducted in-situ, the density of craters remains the main chronometer used for dating surface units [2, 3]. However, this method critically depends on the correct identification of primary craters, as secondary craters (formed by ejecta from a primary impact) and ghost craters (highly degradedor buried) must be excluded to avoid significant overestimations of surface ages [4]. As the identification of crater morphological features is still a long, repetitive, and subjective task when performed manually, the application of modern computer vision techniques has become more and more relevant. While automated crater detection has seen substantial progress in recent years thanks to deep learning and computer vision techniques [5, 6, 7], the classification of craters based on their morphology remains largely unexplored. Yet, such classification is essential to ensure both the validity of crater inventories and the robustness of derived age estimates. Dataset and Preprocessing To train our classifier, we relied on the comprehensive work of Lagain et al. (2021) [4], which provides a manually annotated catalogue of more than 376,000 craters with a size superior at 1km in diameter into four morphological classes: Regular, Secondary, Ghost, and Layered. Image patches centered on each crater are extracted from the global CTX mosaic [8], after reprojection in local stereographic coordinates to preserve the circular geometry of craters at high latitudes. To ensure robustness, we refine the crater locations and sizes using a circle detection algorithm based on the Hough transform [9]. This preprocessing step significantly improves the alignment between craters and image content, a critical requirement for effective supervised learning. In order to train our model, we used 72,000 classified craters, divided in train (28,000 crater), validation (6,000 craters) and test (45,000 craters). Methodology We trained a convolutional neural network classifier based on the YOLOv11 architecture, using a balanced and augmented subset of the crater database. Each image patch is resized and normalized, and we apply standard data augmentation strategies including rotations, flips, and artificial masking to simulate realistic artefacts in CTX images. The model outputs is a classification among the four crater classes describe previously. raining was conducted over 40 epochs on a high-performance multi-GPU server using a cross-entropy loss function and a cosine decayed learning rate schedule. Figure1 show the improvement of accuracy through the learning phase on the validation dataset.Figure1: Validation accuracy with respect to learning epochs Results The final model achieves a classification accuracy of over 80% on a geographically diverse and independent test subdataset containing over 45,000 craters. The Figure2 shows the confusion matrix which gaves us a good insight as how the classification model performed. Performance remains consistent across latitudes. Figure3 shows the classification made on 12 example craters, showing excellent classification, including robustness to illumination conditions and image condition (corrupted data). We also demonstrate the practical use of our classification model in the context of surface dating. By comparing cumulative crater size-frequency distributions (CSFD) before and after removing ghost and secondary craters, we show that automated filtering improves the coherence of the inferred ages with those expected from established crater chronologies.Figure2: Confusion matrix made on the test subdataset. These results show for instance that 80% of true Ghost crater—which represent 2703 craters—where correctly classified, and 2% of them (84 instances) where misclassified as Layered. Overall, the performance is excellent. The regular crater appears slightly more difficult to classify, most probably due to human misclassification. Figure3: Example of 12 crater present in a test area between -100° and -92°E longitude and 0° to 8°S latitude, which were, from top to bottom, classified as Ghost, Layered and Regular. Discussion and ConclusionWe present a novel, scalable, and accurate pipeline for automatic crater classification, which complements existing detection models and provides a new tool for planetary surface dating. This study represents the first fully automated morphological classification of Martian impact craters using deep learning. Our results demonstrate the potential of AI-based approaches to improve crater-based chronostratigraphy, especially when applied systematically to global datasets.As a future work, we plan to extend the model to the Moon and Mercury using transfer learning, but also incorporate additional crater classes or features (e.g., central peaks, double-layer ejecta). Finally, the plan to refine existing Martian chronologies using the filtered crater populations. References[1] W. K. Hartmann, G. Neukum, Cratering chronology and the evolution ofmars, Space Science Reviews 96 (2001) 165–194.[2] G. Neukum, B. Ivanov, W. Hartmann, Cratering records in the inner solar system in relation to the lunar reference system (2001).[3] B. A. Ivanov, Mars/moon cratering rate ratio estimates, Chronology and Evolution of Mars 87, 2001.[4] A. Lagain, S. Bouley, & al., Mars crater and database: A and participative project for the classification of and the morphological characteristics of large martian and craters, The Geological, 2021.[5] G. K. Benedix, A. Lagain, K. Chai, S. Meka, S. Anderson, C. Norman, P. A. Bland, J. Paxman, M. C. Towner, T. Tan, Deriving surface ages on mars using automated crater counting, 2020.[6] R. La Grassa, G. Cremonese, I. Gallo, C. Re, E. Martellato, Yololens: A deep learning model based on super-resolution to enhance the crater...

2 July 2026
Verified
Conference Paper
Open access
MLMyriam Lemelin
Myriam Lemelin
LMLéonard Martinez
Léonard Martinez
GCGuilhem Calas
Guilhem Calas
ALAbigaëlle Lagneaux
Abigaëlle Lagneaux
MRMathieu Roy
Mathieu Roy
DBDavid T. Blewett
David T. Blewett
IntroductionLunar Vertex, NASA’s first Payloads and Research Investigations on the Surface of the Moon (PRISM) delivery, will explore the Reiner Gamma lunar swirl (Figure 1) (Blewett et al., 2025). Swirls consist of high and low albedo variations in the form of loops and ribbons sometimes associated with crustal magnetic “anomalies”. Different hypotheses have been proposed to explain the presence of swirls (summarized in Blewett et al., 2021), including magnetic shielding of the surface from the darkening effects of the solar wind, scouring by gas and dust during the collision of a comet's coma or a meteoroid swarm, and accumulation of fine-grained dust in response to magnetic or electric fields associated with the magnetic anomaly. Studies have investigated the photometric properties of the uppermost regolith at swirls without reaching a consensus regarding their origin but suggest that solar wind shielding and dust levitation/sorting may both play a role (e.g., Denevi et al., 2016; Chrbolková et al. 2019; Kinczyk et al., 2025). Geomorphological studies suggest that on-swirl regions are lower than the off-swirl region by ~4 m at Reiner Gamma (Weirich et al., 2023), supporting the dust migration hypothesis, or local topographic variations pre-dating the formation of the swirls.In this study, we investigate potential differences in regolith properties between on-swirl and off-swirl regions at Reiner Gamma to provide context for the Lunar Vertex mission. Our first objective is to quantify morphometric parameters from the bowl-shaped crater population to investigate potential dust migration mechanisms which may contribute to preferential crater infill on-swirl, or different regolith properties. Our second objective is to estimate regolith thickness using the non-bowl-shaped crater population to see if variations in local regolith thicknesses exist, and in turn, if dust migration mechanisms or differences in pre-impact topography could contribute to these variations.Datasets and methodsWe trained and used a YOLOv5-based model (e.g., Bickel et al., 2025) to detect craters across Reiner Gamma in hillshade images derived from Lunar Reconnaissance Orbiter (LRO) Narrow Angle Camera (NAC) digital terrain models (DTMs) at spatial resolutions from 2 to 5 meters. We developed “MorphoPy”, a Python-based automated crater morphology characterization tool (Martinez et al., 2026), which calculates morphometric parameters (e.g., depth, diameter, wall slope, rim slope, circularity, eccentricity, freshness) on DTMs for craters that have 36 semi-profiles each containing 10 valid measurements. Morphometric parameters are calculated for craters with a diameter (D) equal to or larger than 20 times the spatial resolution of the input DTM, here D≥40 m. In parallel, we manually identified non-bowl-shaped craters (e.g., flat floored, concentric craters, mound) in LROC NAC images on Quickmap and estimated regolith thicknesses using the formula of Bart et al. (2011).Figure 1. Map of Reiner Gamma showing the location of NAC DTMs used to investigate crater morphology overlain by the intensity of the magnetic field (Ravat et al., 2020) (swirl outline from Denevi et al., 2016).Preliminary resultsThe YOLOv5 algorithm identified ~70 000 craters, and “MorphoPy” calculated morphometric parameters for ~25 000 of those. To investigate a subset of the freshest primary craters, we extracted values for craters having a circularity ≥ 0.9, opposite rim slope < 8˚ and a measurable rim for ≥ 50% of their half-profiles. Plots of depth versus diameter values (Figure 2) suggest that most of the freshest primary craters in the Reiner Gamma region have d/D values centered around ~0.1, especially for strength-dominated craters (D < 400 m). Depth versus diameter trends suggest that off-swirl craters are systematically deeper (hence have higher d/D values) than their on-swirl counterpart for a given crater diameter. Analyses will be conducted next to see if the trends are statistically different.Figure 2. Depth versus diameter values for the freshest on-swirl and off-swirl craters in DTMs of different spatial resolution (solid line d/D=0.2, dashed line d/D=0.1, dotted line d/D=0.05).We identified 918 non-bowl-shaped craters on Quickmap, and calculated regolith thickness for the 503 craters that exhibit a clear rim (apparent diameter, DA) and “inner feature” (measured at the base of the normal crater wall slope, DF). Bart et al. (2014) suggest that craters where DF/DA varies between 0.2 and 0.7 and where DA < 300 m are well suited to infer regolith thickness. In our case, 467 craters respect such values (296 off swirl, 171 on swirl) and yield a mean regolith thickness of 6.6 m off-swirl versus 6.1 on-swirl and median values of respectively 4.1 and 4.2 m. These values are consistent with those reported for mare surfaces (3.1–7.8 m; Cooper et al., 1974; Nakamura et al., 1975; Bart et al., 2011). The “regolith thickness” values calculated range between ~1 and 50 m. Impact craters that encounter a strength transition in the target have non-bowl-shaped morphologies. The strength transition can be a layer of regolith over bedrock but can also be a strength transition farther beneath the surface such as layering in the basalt (Bart et al, 2014). A hot spot analysis (Getis-Ord Gi) on the 467 regolith thickness measurements suggests that some high (hot spot) and low (cold spot) regolith thickness values cluster spatially and are statistically significant. Hot spots mostly occur on-swirl and towards the southern portion of Reiner Gamma, while cold spots mostly occur off-swirl and towards the northern portion of Reiner Gamma. The region at ~7.5˚N, 59.5˚W notably exhibits cold spots (regolith thickness values on the order of 1-3 m) in proximity to hot spots (“regolith thickness” or strength contrast at 5-50 m).Figure 3....

9 April 2026
Verified
Thesis
Closed access
LMLéonard Martinez
Léonard Martinez
AC/DC : détection et caractérisation automatique de cratères: L'étude des cratères d'impact à la surface des corps telluriques constitue un outil central pour dater les terrains et reconstruire l'histoire géologique des planètes. Cette méthode repose toutefois sur une connaissance fine de la morphologie des cratères, car certaines morphologies, comme les cratères secondaires ou fortement dégradés, doivent être exclus des comptages sous peine de biaiser les résultats. Pendant cette thèse, nous avons développé de nouvelles méthodes de traitement en combinant à la fois des techniques d'apprentissage à base de réseaux de neurone artificiels et des méthodes de traitement d'images et d'analyse géomorphologique avancée. Grâce à cela nous sommes capables de détecter automatiquement les cratères présents dans les images à haute résolution, de corriger leur géométrie, puis de les classifier selon leur morphologie. Nous nous appuyons sur les mosaïques globales de Mars acquises par la caméra CTX et sur une base de données de cratères classifiés, que nous avons améliorées par une méthode de détection de cercles fondée sur la transformée de Hough. Cette étape améliore la précision géométrique des annotations, en recentrant automatiquement les cratères et en ajustant leur rayon au plus proche de la structure visible sur l'image. Nous avons ensuite conçu une méthode de détection multi-échelle permettant d'identifier tous les cratères, quelles que soient leur taille, géométrie, éclairement ou position en latitude / longitude. Chaque image est analysée à différentes résolutions, et les résultats sont fusionnés pour éviter les redondances. Nous avons entraîné un modèle de classification profond (YOLOv11) capable d'attribuer à chaque cratère l'une des quatre classes géomorphologiques : ordinaire, secondaire, fantôme ou à rempart fluidisé. La dernière partie porte sur la détermination de l'ordre de superposition des cratères, lorsqu'ils se recouvrent. L'ensemble constitue une suite d'outils robuste pour automatiser la cartographie géomorphologique planétaire, améliorer la fiabilité des datations et, à terme, faciliter l'analyse d'autres corps. Tous ces outils ont été évalués et validés sur des données totalement indépendantes de l'entrainement pour une utilisation scientifique avec des incertitudes maitrisées. En perspective, cette suite d'outils permettra de mieux appréhender les larges volumes de données de l'imagerie astronomique, car les objets d'intérêts sont souvent présents à plusieurs échelles. En particulier, nous proposons d'étudier la détection et la caractérisation de cratères à la surface des planètes et autres corps du système solaire.

29 August 2026 in Journal of Geophysical Research: Machine Learning and Computation
Verified
Journal Article
Open access
PDF file
LML. Martinez
L. Martinez
FAF. Andrieu
F. Andrieu
FSF. Schmidt
F. Schmidt
MBM. S. Bentley
M. S. Bentley
Identifying and classifying impact craters on Mars is crucial for understanding the planet's geological history and surface evolution. Traditional crater classification relies on manual annotation methods, which are often limited by human biases and the difficulty of interpreting geomorphological features. Classifying crater is both necessary and challenging. This task is necessary because planetary surfaces ages are estimated by crater density, but some crater types have to be discarded from the analysis to avoid large estimation errors. For instance, secondary impact craters or buried craters should not be counted for a given surface unit. This task is challenging because the different crater classes share a lot of common features. In this study, we present a deep‐learning approach for automated crater classification on Mars using a YOLOv11 architecture. Our method is trained on downsampled (50 m/pixel) Context Camera images (CTX camera) and a human‐annotated crater database of craters 1 km in diameter. This pipeline is one of the first automated tools capable of classifying craters 1 km, addressing a critical gap in high‐resolution analyses. Validated across diverse Martian terrains, it shows potential for planetary dating and comparative geology. Incorporating a pre‐processing step to reduce false positives, we train and test our model globally on separate data sets. Our results demonstrate 81% accuracy, providing a reliable tool for large‐scale planetary surface analysis. This pipeline advances planetary geosciences by enabling systematic, feature‐based crater classification and highlights deep learning's potential in planetary exploration.