arXiv subjects

Kento Sasaki

Publications and source records attributed to Kento Sasaki.

At least 19 recordsLinked to original sources

arXiv2026

Engineering Nanodiamonds for Quantum Sensing: Material Constraints at the Nanoscale

Optically addressable solid-state spin defects have emerged as powerful multimodal quantum sensors, with nitrogen-vacancy (NV) centers in bulk diamond providing benchmark quantum control and sensitivity under ambient conditions. Embedding such defects in nanodiamonds (NDs) extends these capabilities to mobile probes capable of accessing complex biological and nanoscale environments. Reduced dimensions, however, introduce constraints beyond volumetric spin impurities, notably enhanced lattice strain and surface-induced noise sources, which shorten NV spin relaxation times (T1 and T2) and destabilize the NV charge state, as well as resulting in pronounced particle-to-particle variability in NDs typically produced by top-down approaches. These effects complicate both sensing performance and the quantitative interpretation of multimodal signals in realistic environments. This article provides a structured perspective on the physical mechanisms by which material properties constrain NV behavior in NDs, together with mitigation strategies that shape the robust use of these mobile quantum sensors for biosensing and nanoscale science.

quant-ph↗

arXiv2026

High-pressure magnetic transition in iron observed via diamond quantum sensing

Diamond quantum sensors offer high precision and spatial resolution as magnetic probes, making them promising for a wide range of applications. While diamond anvil cells (DACs) can generate extremely high pressures, techniques for magnetometry under such conditions remain limited. By fabricating an ensemble of NV centers directly on the anvil diamond surface, we enable precise magnetic measurements under high pressure. In this work, we employ this NV ensemble to image the stray magnetic field of iron up to 30 GPa, enabling the observation of the magnetic transition ($\alpha$-$\varepsilon$ transition) in iron.

physics.app-ph↗

arXiv2026

Spin-coherence characterization of boron vacancy defects in hexagonal boron nitride with broadband microwave pulses

Negatively charged boron vacancy (VB-) defects in hexagonal boron nitride (hBN) are promising for nanoscale-proximity quantum sensing. To evaluate their performance, it is important to characterize the spin coherence times T2* and T2. In this study, we realized sub-GHz Rabi oscillations of VB- using an isotopically enriched hBN thin film directly stamped onto a narrow gold wire. Using these strong microwave pulses, we performed Ramsey interference and Hahn echo measurements. The Ramsey interference signal showed Gaussian-like decay, yielding T2* = 13.8 ns. The Hahn echo measurement gave T2 = 108.7 ns and a stretch factor of {\alpha}= 1.25. These results experimentally clarify the spin coherence properties of VB- and provide an effective method for evaluating the coherence of spin defects in van der Waals thin films with broad resonance linewidths.

cond-mat.mes-hall↗

arXiv2026

Wide-field magnetic imaging of shielding-current-driven vortex rearrangement under local heating using diamond quantum sensors

Understanding and controlling vortex motion in superconductors are important both for suppressing dissipation in superconducting devices and for device applications that exploit vortices. In this work, we quantitatively imaged the stray magnetic field distribution of vortices in an NbN thin film by wide-field magnetic imaging using a perfectly aligned diamond nitrogen-vacancy ensemble. By continuously measuring while stepwise varying the applied magnetic field under local laser heating, we captured a rearrangement of the vortex configuration in real space and in real time over more than 100 min. The observed vortex rearrangement is consistent with a reduction of the pinning force due to local laser heating and with the Lorentz force exerted by shielding currents induced by the field variation. These results provide insight into vortex dynamics and suggest potential applications, including vortex exclusion from sensitive regions of superconducting devices.

cond-mat.supr-con↗

arXiv2025

Read or Ignore? A Unified Benchmark for Typographic-Attack Robustness and Text Recognition in Vision-Language Models

Large vision-language models (LVLMs) are vulnerable to typographic attacks, where misleading text inserted into an image can override visual understanding. However, existing evaluation protocols and defenses are largely focused on object recognition and do not consider text-reading capability. This is a critical oversight: real-world scenarios often require both recognizing objects and reading scene text (e.g., recognizing pedestrians while reading traffic signs), where simply ignoring all text for robustness is unacceptable in practice. To address this gap, we introduce a novel task, Read-or-Ignore VQA (RIO-VQA), which jointly evaluates both requirements: models must decide, from context, when to read scene text and when to ignore inserted distractor text. To evaluate this capability, we present RIO-Bench, a same-scene counterfactual benchmark that holds the scene fixed while varying only question intent (object vs. text) and text condition (clean vs. attack), enabling direct comparisons of model behaviors with reduced confounding factors. Using RIO-Bench, we highlight a trade-off: representative defenses developed in object-centric settings can achieve robustness by suppressing text sensitivity, at the cost of text-reading performance (i.e., "ignoring" text). Motivated by this trade-off, we provide a data-driven defense baseline that improves both requirements on RIO-Bench, complementing prior text-ignoring baselines. Overall, this work highlights a fundamental misalignment between the current object-centric robustness scope and real-world multimodal requirements, providing a principled path toward reliable LVLMs.

cs.CV↗

arXiv2025

Observation of individual vortex penetration in a coplanar superconducting resonator

We demonstrate the detection and control of individual Abrikosov vortices in superconducting microwave resonators. $\lambda/4$ resonators with a narrowed region near the grounded end acting as a vortex trap were fabricated and studied using microwave transmission spectroscopy at millikelvin temperatures. Sharp stepwise drops in resonance frequency are detected as a function of increasing external magnetic field, attributed to the entry of individual Abrikosov vortices in the narrow region. This interpretation is confirmed by NV center magnetometry revealing discrete vortex entry events on increasing field. Our results establish a method to investigate and manipulate the states of Abrikosov vortices with microwaves.

cond-mat.supr-con↗

arXiv2025

Visualization of Current-Driven Vortex Formation in High-$T_c$ Cuprate Superconductors

Type-II superconductors exhibit hysteretic behavior due to the presence of quantum vortices, and the order in which temperature and external field are varied plays a decisive role. Here we take current, rather than magnetic field, as the external drive. We image the magnetic field of a high-$T_c$ cuprate superconductor strip after cooling. We confirm that even in zero magnetic field, current-biased cooling nucleates vortices within the strip. With a small external magnetic field, the distribution is polarized opposite to the Lorentz-force direction. These behaviors follow from the self-consistent relation between current and local field in steady flux flow. Our findings show that current history is encoded as vortices. This reveals self-field effects that influence dc measurements and glassy transitions under drive.

cond-mat.supr-con↗

arXiv2025

Hierarchical Reasoning with Vision-Language Models for Incident Reports from Dashcam Videos

Recent advances in end-to-end (E2E) autonomous driving have been enabled by training on diverse large-scale driving datasets, yet autonomous driving models still struggle in out-of-distribution (OOD) scenarios. The COOOL benchmark targets this gap by encouraging hazard understanding beyond closed taxonomies, and the 2COOOL challenge extends it to generating human-interpretable incident reports. We present a hierarchical reasoning framework for incident report generation from dashcam videos that integrates frame-level captioning, incident frame detection, and fine-grained reasoning within vision-language models (VLMs). We further improve factual accuracy and readability through model ensembling and a Blind A/B Scoring selection protocol. On the official 2COOOL open leaderboard, our method ranks 2nd among 29 teams and achieves the best CIDEr-D score, producing accurate and coherent incident narratives. These results indicate that hierarchical reasoning with VLMs is a promising direction for accident analysis and for broader understanding of safety-critical traffic events. The implementation and code are available at https://github.com/riron1206/kaggle-2COOOL-2nd-Place-Solution.

cs.CV↗

arXiv2025

Understanding Sensitivity of Differential Attention through the Lens of Adversarial Robustness

Differential Attention (DA) has been proposed as a refinement to standard attention, suppressing redundant or noisy context through a subtractive structure and thereby reducing contextual hallucination. While this design sharpens task-relevant focus, we show that it also introduces a structural fragility under adversarial perturbations. Our theoretical analysis identifies negative gradient alignment-a configuration encouraged by DA's subtraction-as the key driver of sensitivity amplification, leading to increased gradient norms and elevated local Lipschitz constants. We empirically validate this Fragile Principle through systematic experiments on ViT/DiffViT and evaluations of pretrained CLIP/DiffCLIP, spanning five datasets in total. These results demonstrate higher attack success rates, frequent gradient opposition, and stronger local sensitivity compared to standard attention. Furthermore, depth-dependent experiments reveal a robustness crossover: stacking DA layers attenuates small perturbations via depth-dependent noise cancellation, though this protection fades under larger attack budgets. Overall, our findings uncover a fundamental trade-off: DA improves discriminative focus on clean inputs but increases adversarial vulnerability, underscoring the need to jointly design for selectivity and robustness in future attention mechanisms.

cs.LG↗

arXiv2025

STRIDE-QA: Visual Question Answering Dataset for Spatiotemporal Reasoning in Urban Driving Scenes

Vision-Language Models (VLMs) have been applied to autonomous driving to support decision-making in complex real-world scenarios. However, their training on static, web-sourced image-text pairs fundamentally limits the precise spatiotemporal reasoning required to understand and predict dynamic traffic scenes. We address this critical gap with STRIDE-QA, a large-scale visual question answering (VQA) dataset for physically grounded reasoning from an ego-centric perspective. Constructed from 100 hours of multi-sensor driving data in Tokyo, capturing diverse and challenging conditions, STRIDE-QA is the largest VQA dataset for spatiotemporal reasoning in urban driving, offering 16M QA pairs over 270K frames. Grounded by dense, automatically generated annotations including 3D bounding boxes, segmentation masks, and multi-object tracks, the dataset uniquely supports both object-centric and ego-centric reasoning through three novel QA tasks that require spatial localization and temporal prediction. Our benchmarks demonstrate that existing VLMs struggle significantly, with near-zero scores on prediction consistency. In contrast, VLMs fine-tuned on STRIDE-QA exhibit dramatic performance gains, achieving 55% success in spatial localization and 28% consistency in future motion prediction, compared to near-zero scores from general-purpose VLMs. Therefore, STRIDE-QA establishes a comprehensive foundation for developing more reliable VLMs for safety-critical autonomous systems.

cs.CV↗

arXiv2025

Beyond ensemble averaging: Parallelized single-shot readout of hole capture in diamond

Understanding the generation, transport and capture of charge carriers in semiconductors is of fundamental technological importance. However, the ensemble measurement techniques ubiquitous in electronics offer limited insight into the nanoscale environment that is crucial to the operation of modern quantum-electronic devices. Here, we combine widefield optical microscopy with precision spectroscopy to examine the capture of photogenerated holes by negatively charged nitrogen vacancy (NV-) centers in diamond. Simultaneous single-shot charge readout over hundreds of individual NVs allows us to resolve the roles of ionized impurities, reveal the formation of space charges fields, and monitor the thermalization of hot photo-carriers during diffusion. We measure effective NV- hole capture radii in excess of 0.2 um, a value approaching the Onsager limit and made possible here thanks to the near-complete neutralization of coexisting charge traps. These results establish a new platform for resolving charge dynamics beyond ensemble averages, with direct relevance to nanoscale electronics and quantum devices.

cond-mat.mes-hall↗

arXiv2025

GPa Pressure Imaging Using Nanodiamond Quantum Sensors

We demonstrate wide-field optical microscopy of the pressure distribution at approximately 20 GPa in a diamond anvil cell (DAC), using nitrogen-vacancy (NV) centers in nanodiamonds (NDs) as quantum sensors. Pressure and non-hydrostaticity maps are obtained by fitting optically detected magnetic resonance (ODMR) spectra with models incorporating hydrostatic and uniaxial stress conditions. Two methods for introducing NDs with a pressure-transmitting medium are compared, revealing that the embedding approach affects the degree of non-hydrostaticity. This ND-based technique offers a powerful imaging platform for probing pressure-induced phenomena and is extendable to other physical quantities such as magnetic fields.

cond-mat.mtrl-sci↗

arXiv2025

Nanodiamond quantum thermometry assisted with machine learning

Nanodiamonds (NDs) are quantum sensors that enable local temperature measurements, taking advantage of their small size. Though the model based analysis methods have been used for ND quantum thermometry, their accuracy has yet to be thoroughly investigated. Here, we apply model-free machine learning with the Gaussian process regression (GPR) to ND quantum thermometry and compare its capabilities with the existing methods. We prove that GPR provides more robust results than them, even for a small number of data points and regardless of the data acquisition methods. This study extends the range of applications of ND quantum thermometry with machine learning.

quant-ph↗

arXiv2025

Quantitative imaging of nonlinear spin-wave propagation using diamond quantum sensors

Spin waves propagating in magnetic materials exhibit nonlinear behavior at large amplitudes due to the competition between excitation and relaxation, providing an attractive platform for exploring nonlinear wave dynamics. In particular, spin waves with a non-zero wavenumber that carry momentum undergo nonlinear relaxation and experience wavenumber modulation in the nonlinear regime. This nonlinearity has been observed experimentally, for example in S. R. Lake et al., Phys. Rev. Appl. 17, 034010 (2022), but a quantitative comparison with theory has not yet been carried out. Here, We image nonlinear spin-wave propagation in two yttrium iron garnet thin films with distinct spin-wave decay rates using a wide-field quantum diamond microscope. We obtain quantitative distributions of spin-wave amplitude and phase as a function of the excitation microwave strength. As a result, we observe a threshold in the spin-wave amplitude beyond which nonlinear effects become evident and confirm that this threshold is consistent with theoretical predictions based on four-magnon scattering for both samples. Moreover, as the amplitude of the spin waves increases, we observe modulation of the wavenumber across the field of view. We attribute this modulation primarily to a reduction in the saturation magnetization caused by incoherent spin waves generated by multi-magnon scattering. Our quantitative measurements provide a pathway for visualizing nonlinear spin-wave dynamics and are crucial for deepening our understanding of the underlying mechanisms.

cond-mat.mes-hall↗

arXiv2025

Systematic investigation of dynamic nuclear polarization with boron vacancy in hexagonal boron nitride

Dynamic nuclear polarization (DNP) using the boron vacancy ($\mathrm{V_B^-}$) in hexagonal boron nitride (hBN) has gained increasing attention. Understanding this DNP requires systematically investigating the optically detected magnetic resonance (ODMR) spectra and developing a model that quantitatively describes its behavior. Here, we measure the ODMR spectra of $\mathrm{V_B^-}$ in $\mathrm{h}^{10}\mathrm{B}^{15}\mathrm{N}$ over a wide magnetic field range, including the ground state level anti-crossing (GSLAC), and compare them with the results of the Lindblad-based simulation that considers a single electron spin and three neighboring $^{15}\mathrm{N}$ nuclear spins. Our simulation successfully reproduces the experimental spectra, including the vicinity of GSLAC. It can explain the overall behavior of the magnetic field dependence of the nuclear spin polarization estimated using the Lorentzian fitting of the spectra. Despite such qualitative agreement, we also demonstrate that the fitting methods cannot give accurate polarizations. Finally, we discuss that symmetry-induced mechanisms of $\mathrm{V_B^-}$ limit the maximum polarization. Our study is an essential step toward a quantitative understanding of DNP using defects in hBN and its quantum applications.

cond-mat.mes-hall↗

arXiv2025

One-D-Piece: Image Tokenizer Meets Quality-Controllable Compression

Current image tokenization methods require a large number of tokens to capture the information contained within images. Although the amount of information varies across images, most image tokenizers only support fixed-length tokenization, leading to inefficiency in token allocation. In this study, we introduce One-D-Piece, a discrete image tokenizer designed for variable-length tokenization, achieving quality-controllable mechanism. To enable variable compression rate, we introduce a simple but effective regularization mechanism named "Tail Token Drop" into discrete one-dimensional image tokenizers. This method encourages critical information to concentrate at the head of the token sequence, enabling support of variadic tokenization, while preserving state-of-the-art reconstruction quality. We evaluate our tokenizer across multiple reconstruction quality metrics and find that it delivers significantly better perceptual quality than existing quality-controllable compression methods, including JPEG and WebP, at smaller byte sizes. Furthermore, we assess our tokenizer on various downstream computer vision tasks, including image classification, object detection, semantic segmentation, and depth estimation, confirming its adaptability to numerous applications compared to other variable-rate methods. Our approach demonstrates the versatility of variable-length discrete image tokenization, establishing a new paradigm in both compression efficiency and reconstruction performance. Finally, we validate the effectiveness of tail token drop via detailed analysis of tokenizers.

cs.CV↗

arXiv2024

Wideband wide-field imaging of spin-wave propagation using diamond quantum sensors

Imaging spin-wave propagation in magnetic materials in a wide frequency range is crucial for understanding and applying spin-wave dynamics. Recently, nitrogen-vacancy (NV) centers in diamond have attracted attention as sensors capable of quantitatively measuring the amplitude and phase of coherent spin waves. However, the conventional sensing protocol has been limited to detecting spin waves whose frequencies match the resonance frequency of the NV spins. We demonstrate that by utilizing the AC Zeeman effect, it is possible to image spin waves propagating in a yttrium iron garnet (YIG) thin film over a wide frequency range up to a maximum detuning of 567 MHz without changing the external magnetic field. Our results expand the applicability of NV centers for spin-wave sensing and pave the way for quantitative investigations of the dynamics in various magnetic materials, such as metallic ferromagnets and van der Waals magnets.

cond-mat.mes-hall↗

arXiv2024

Systematic characterization of nanoscale $h$-BN quantum sensor spots created by helium-ion microscopy

The nanosized boron vacancy ($V_\mathrm{B}^-$) defect spot in hexagonal boron nitride ($h$-BN) is promising for a local magnetic field quantum sensor. One of its advantages is that a helium-ion microscope can make a spot at any location in an $h$-BN flake with nanometer accuracy. In this study, we investigate the properties of the created nanosized $V_\mathrm{B}^-$ defect spots by systematically varying three conditions: the helium-ion dose, the thickness of the $h$-BN flakes, and the substrate on which the $h$-BN flakes are attached. The physical background of the results obtained is successfully interpreted using Monte Carlo calculations. From the findings obtained here, a guideline for their optimal creation conditions is obtained to maximize its performance as a quantum sensor concerning sensitivity and localization.

cond-mat.mes-hall↗