Lukáš Adam
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Lukáš Adam has a total of 99 co-authors. Here are the top co-authors by number of shared publications:
- Kostas Papafitsoros
Queen Mary University of London
10 shared publications - Lukáš Picek
University of West Bohemia
8 shared publications - Václav Šmídl
Faculty of Electrical Engineering, Czech Technical University in Prague, Prague, Czech Republic
6 shared publications - Alexis Joly
Inria LIRMM Univ Montpellier CNRS Montpellier France
4 shared publications - Hervé Goëau
AMAP, Université de Montpellier, CIRAD, INRAE, IRD
4 shared publications - Holger Klinck
K. Lisa Yang Center for Conservation Bioacoustics Cornell University Ithaca New York USA
4 shared publications
Latest Publications
- Open Access
A perturbation atlas of mouse neural organoids resolves macroglial reactivity and nominates drivers of degeneration
Understanding neuro-glial roles in homeostasis and disease requires scalable central nervous system (CNS) models, particularly as macroglia like astrocytes and Müller glia remain challenging to study in slow-developing human systems. Here, we generate and map the cellular diversity of mouse neural organoids spanning multiple brain regions and the retina. Benchmarking transcriptomic maturation against in vivo references captures postnatal neuromacroglial states within a month. We show that macroglia are highly responsive to inflammatory cues and adopt specific transcriptomic states also found in vivo . Through a single-cell perturbation screen targeting 13 Müller-glia-specific pathways, we map how certain inflammatory cytokines drive degeneration-like stress programs in photoreceptors and distinct reactive states in Müller glia, including an Oncostatin M-driven chaperone-associated state. Mechanistically, we identify gp130-STAT3 signaling as a targetable driver of macroglial reactivity and photoreceptor distress. Together, these single-cell atlases provide a foundation for dissecting cellular crosstalk and neuro-glial pathology in the mature CNS.
Open AccessA perturbation atlas of mouse neural organoids resolves macroglial reactivity and nominates drivers of degeneration
Understanding neuro-glial roles in homeostasis and disease requires scalable central nervous system (CNS) models, particularly as macroglia like astrocytes and Müller glia remain challenging to study in slow-developing human systems. Here, we generate and map the cellular diversity of mouse neural organoids spanning multiple brain regions and the retina. Benchmarking transcriptomic maturation against in vivo references captures postnatal neuromacroglial states within a month. We show that macroglia are highly responsive to inflammatory cues and adopt specific transcriptomic states also found in vivo . Through a single-cell perturbation screen targeting 13 Müller-glia-specific pathways, we map how certain inflammatory cytokines drive degeneration-like stress programs in photoreceptors and distinct reactive states in Müller glia, including an Oncostatin M-driven chaperone-associated state. Mechanistically, we identify gp130-STAT3 signaling as a targetable driver of macroglial reactivity and photoreceptor distress. Together, these single-cell atlases provide a foundation for dissecting cellular crosstalk and neuro-glial pathology in the mature CNS.
- 21 September 2026
- Published by Springer Nature
- p. 519-547
Overview of LifeCLEF 2026: AI Challenges for Biodiversity Understanding and Ecosystem Management
- 21 September 2026
- Published by Springer Nature
- p. 519-547
Overview of LifeCLEF 2026: AI Challenges for Biodiversity Understanding and Ecosystem Management
- Open Access
Automated wildlife re-identification by merging information from multiple body parts: A case study in sea turtles
Wildlife re-identification (re-ID) is a widely used and powerful tool with diverse applications in animal ecology and conservation. Current automated methods typically operate on single images of a single body part of the animal. However, a single encounter may contain multiple images capturing different body regions, each providing complementary individual-specific information. In contrast to automated approaches, researchers often manually select the most suitable images and regions for identification based on factors like visibility, occlusion and image quality. This creates a mismatch between automated methods and field practice, limiting the practical adoption of current automated re-ID pipelines. Here, we address this by introducing an encounter-based, multi-body-part re-ID framework, using sea turtles as a model taxon. Our framework combines three elements: (1) An orientation-aware deep learning model, TurtleDetector, that in addition to the full bodies, it also automatically segments key body regions, i.e. heads, front and hind flippers, from images within an encounter; (2) a hybrid body-part-specific retrieval method, that sequentially combines a fast global-feature model (MiewID or DINOv3) with a more accurate but costlier local-feature model (ALIKED with LightGlue); and (3) a merged identity-prediction strategy that selects the highest calibrated similarity score across all available body parts and images of an encounter. We evaluate the framework on three long-term re-ID datasets spanning three species, loggerheads, greens, and hawksbill turtles, under an evaluation protocol that mirrors real-world, time-aware re-ID workflows. Across datasets, combining multiple body regions consistently improved identification performance over the best-performing single body region, resulting to an increase of 4–6% in top-1 accuracy. Interestingly, body regions traditionally underused in sea turtle re-ID, such as the hind flippers and carapaces, provided complementary identifying information that improved encounter-level re-ID when integrated through the hybrid retrieval method. Our findings demonstrate that automated wildlife re-ID can benefit from moving beyond single-image, single-body-part identification towards encounter-level integration of all available visual evidence. Our work further suggests that, where feasible, field photo-acquisition protocols should aim to capture multiple informative views of an individual during each encounter. Importantly, many species and taxa, including elephants, primates, cetaceans, and other large vertebrates, possess such individual-specific features across multiple body regions, highlighting the broad potential applicability of our framework.
Open AccessAutomated wildlife re-identification by merging information from multiple body parts: A case study in sea turtles
Wildlife re-identification (re-ID) is a widely used and powerful tool with diverse applications in animal ecology and conservation. Current automated methods typically operate on single images of a single body part of the animal. However, a single encounter may contain multiple images capturing different body regions, each providing complementary individual-specific information. In contrast to automated approaches, researchers often manually select the most suitable images and regions for identification based on factors like visibility, occlusion and image quality. This creates a mismatch between automated methods and field practice, limiting the practical adoption of current automated re-ID pipelines. Here, we address this by introducing an encounter-based, multi-body-part re-ID framework, using sea turtles as a model taxon. Our framework combines three elements: (1) An orientation-aware deep learning model, TurtleDetector, that in addition to the full bodies, it also automatically segments key body regions, i.e. heads, front and hind flippers, from images within an encounter; (2) a hybrid body-part-specific retrieval method, that sequentially combines a fast global-feature model (MiewID or DINOv3) with a more accurate but costlier local-feature model (ALIKED with LightGlue); and (3) a merged identity-prediction strategy that selects the highest calibrated similarity score across all available body parts and images of an encounter. We evaluate the framework on three long-term re-ID datasets spanning three species, loggerheads, greens, and hawksbill turtles, under an evaluation protocol that mirrors real-world, time-aware re-ID workflows. Across datasets, combining multiple body regions consistently improved identification performance over the best-performing single body region, resulting to an increase of 4–6% in top-1 accuracy. Interestingly, body regions traditionally underused in sea turtle re-ID, such as the hind flippers and carapaces, provided complementary identifying information that improved encounter-level re-ID when integrated through the hybrid retrieval method. Our findings demonstrate that automated wildlife re-ID can benefit from moving beyond single-image, single-body-part identification towards encounter-level integration of all available visual evidence. Our work further suggests that, where feasible, field photo-acquisition protocols should aim to capture multiple informative views of an individual during each encounter. Importantly, many species and taxa, including elephants, primates, cetaceans, and other large vertebrates, possess such individual-specific features across multiple body regions, highlighting the broad potential applicability of our framework.
- 24 March 2026
- Published by Springer Nature
- p. 287-296
LifeCLEF 2026 Teaser: AI Challenges for Biodiversity Understanding and Ecosystem Management
- 24 March 2026
- Published by Springer Nature
- p. 287-296
LifeCLEF 2026 Teaser: AI Challenges for Biodiversity Understanding and Ecosystem Management
- Open Access
- 5 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (3) , 3602-3613
Analysis and Taxonomy of Optimal Current Setpoints for Five-Phase Synchronous Motor Drives
The optimal current setpoints for three-phase machines are known to be divided into operating regions of smooth solutions, such as the maximum torque per ampere (MTPA) and the field weakening (FW) regions. These regions are switched at critical points, such as reaching the maximum of the voltage waveform. This article presents a hypothesis that a general solution of the current setpoints of multiphase machines has the same form. To support this claim, optimal setpoints of a five-phase machine are computed numerically and analyzed from the perspective of constrained optimization theory. A taxonomy of smooth regions generalizing MTPA and FW to five-phase machines is identified. The number of regions of a five-phase machine increases to 9 due to a higher number of constraints arising from more complicated waveforms of phase current and voltage. Moreover, a polynomial approximation of the smooth regions as well as their boundaries is developed and shown to have low approximation error. It is applied to represent optimal setpoints as a function of the requested torque and rotational speed; however, it naturally extends to other parameters of the machine. This is demonstrated by a sensitivity study of the value of the permanent magnet flux and injection of a zero-sequence component (ZSC) of the voltage vector. The theory is validated experimentally on a laboratory prototype of a five-phase PMSM drive of 3 kW.
- 5 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (3) , 3602-3613
Open AccessAnalysis and Taxonomy of Optimal Current Setpoints for Five-Phase Synchronous Motor Drives
The optimal current setpoints for three-phase machines are known to be divided into operating regions of smooth solutions, such as the maximum torque per ampere (MTPA) and the field weakening (FW) regions. These regions are switched at critical points, such as reaching the maximum of the voltage waveform. This article presents a hypothesis that a general solution of the current setpoints of multiphase machines has the same form. To support this claim, optimal setpoints of a five-phase machine are computed numerically and analyzed from the perspective of constrained optimization theory. A taxonomy of smooth regions generalizing MTPA and FW to five-phase machines is identified. The number of regions of a five-phase machine increases to 9 due to a higher number of constraints arising from more complicated waveforms of phase current and voltage. Moreover, a polynomial approximation of the smooth regions as well as their boundaries is developed and shown to have low approximation error. It is applied to represent optimal setpoints as a function of the requested torque and rotational speed; however, it naturally extends to other parameters of the machine. This is demonstrated by a sensitivity study of the value of the permanent magnet flux and injection of a zero-sequence component (ZSC) of the voltage vector. The theory is validated experimentally on a laboratory prototype of a five-phase PMSM drive of 3 kW.
- 3 September 2025
- Published by Springer Nature
- p. 338-362
Overview of LifeCLEF 2025: Challenges on Species Presence Prediction and Identification, and Individual Animal Identification
- 3 September 2025
- Published by Springer Nature
- p. 338-362
Overview of LifeCLEF 2025: Challenges on Species Presence Prediction and Identification, and Individual Animal Identification
- 12 May 2025
- Published by Springer Nature
- p. 18-36
WildFusion: Individual Animal Identification with Calibrated Similarity Fusion
- 12 May 2025
- Published by Springer Nature
- p. 18-36
WildFusion: Individual Animal Identification with Calibrated Similarity Fusion
- Open Access
- 8 May 2025
- Published by Elsevier in Ecological Informatics
- Vol. 89
Exploiting facial side similarities to improve AI-driven sea turtle photo-identification systems
- 8 May 2025
- Published by Elsevier in Ecological Informatics
- Vol. 89
Open AccessExploiting facial side similarities to improve AI-driven sea turtle photo-identification systems
- 3 April 2025
- Published by Springer Nature
- p. 373-381
LifeCLEF 2025 Teaser: Challenges on Species Presence Prediction and Identification, and Individual Animal Identification
- 3 April 2025
- Published by Springer Nature
- p. 373-381
LifeCLEF 2025 Teaser: Challenges on Species Presence Prediction and Identification, and Individual Animal Identification
- 12 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 72 (9) , 8747-8756
Maximizing Output Voltage of Three-Phase Nine-Leg Multilevel STATCOM by Minimum Infinity Norm Optimization
Power converters based on classic three-phase three-leg voltage source inverters (VSIs) that provide a multilevel output voltage are an interesting alternative to multilevel converters that suffer from complexity and higher costs. However, the multiple degrees of freedom of these systems pose a challenge to their optimal control. In this article, the minimum infinity norm algorithm is proposed and implemented for triple three-leg VSI static synchronous compensators (STATCOM) connected to a three-phase power grid. It is shown that the method provides optimal utilization of the dc-link voltage and allows to increase the output voltage by 39% compared to the minimum $\boldsymbol{l}_{\mathbf{2}}$ norm method and by 16% compared to the phase-disposition pulse width modulation (PD-PWM) while keeping total harmonic distortion (THD) below 0.5%. In addition, the algorithm has been optimized for real-time systems, resulting in a very fast algorithm with a worst-case complexity of 39 scalar–scalar multiplications and no division.
- 12 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 72 (9) , 8747-8756
Maximizing Output Voltage of Three-Phase Nine-Leg Multilevel STATCOM by Minimum Infinity Norm Optimization
Power converters based on classic three-phase three-leg voltage source inverters (VSIs) that provide a multilevel output voltage are an interesting alternative to multilevel converters that suffer from complexity and higher costs. However, the multiple degrees of freedom of these systems pose a challenge to their optimal control. In this article, the minimum infinity norm algorithm is proposed and implemented for triple three-leg VSI static synchronous compensators (STATCOM) connected to a three-phase power grid. It is shown that the method provides optimal utilization of the dc-link voltage and allows to increase the output voltage by 39% compared to the minimum $\boldsymbol{l}_{\mathbf{2}}$ norm method and by 16% compared to the phase-disposition pulse width modulation (PD-PWM) while keeping total harmonic distortion (THD) below 0.5%. In addition, the algorithm has been optimized for real-time systems, resulting in a very fast algorithm with a worst-case complexity of 39 scalar–scalar multiplications and no division.