Search arXiv⌕ Search

arXiv subjects

Satoshi Hashimoto

Publications and source records attributed to Satoshi Hashimoto.

8 recordsLinked to original sources
arXiv2026

MuST-VAD: Mutual Structured Learning for Video Anomaly Detection

In this paper, we propose MuST-VAD, a mutual structured learning framework for weakly supervised video anomaly detection (VAD) in which an anomaly detector and a large vision-language model (LVLM) exchange their acquired knowledge. Detectors in weakly supervised VAD learn anomaly scores from features extracted by a fixed, task-agnostic backbone. These fixed features bound the achievable detection accuracy. Recent methods therefore transfer LVLM semantics into the detector as richer features. However, this transfer is one-way: what the detector learns about the target videos never returns to the LVLM. MuST-VAD extends the one-way transfer into a bidirectional learning loop. In this loop, the latest detector predictions supervise the LVLM adaptation, and the adapted LVLM returns updated representations that retrain the detector; the two models alternate these updates over small video groups. Both models train on detector-selected key clips, while confidence weighting and annotation-anchored question answering keep the exchanged supervision reliable. On UCF-Crime, our mutual learning improves the one-pass transfer baseline from 88.15% to 88.63% AUROC and from 37.25% to 42.46% average precision (AP), outperforming the state-of-the-art method in AP by 4.13 points.

cs.CV↗
arXiv2025

CADE: Continual Weakly-supervised Video Anomaly Detection with Ensembles

Video anomaly detection (VAD) has long been studied as a crucial problem in public security and crime prevention. In recent years, weakly-supervised VAD (WVAD) have attracted considerable attention due to their easy annotation process and promising research results. While existing WVAD methods tackle mainly on static datasets, the possibility that the domain of data can vary has been neglected. To adapt such domain-shift, the continual learning (CL) perspective is required because otherwise additional training only with new coming data could easily cause performance degradation for previous data, i.e., forgetting. Therefore, we propose a brand-new approach, called Continual Anomaly Detection with Ensembles (CADE) that is the first work combining CL and WVAD viewpoints. Specifically, CADE uses the Dual-Generator(DG) to address data imbalance and label uncertainty in WVAD. We also found that forgetting exacerbates the "incompleteness'' where the model becomes biased towards certain anomaly modes, leading to missed detections of various anomalies. To address this, we propose to ensemble Multi-Discriminator (MD) that capture missed anomalies in past scenes due to forgetting, using multiple models. Extensive experiments show that CADE significantly outperforms existing VAD methods on the common multi-scene VAD datasets, such as ShanghaiTech and Charlotte Anomaly datasets.

cs.CV↗
arXiv2025

PA-VAD: Diffusion-Based Pseudo-Only Video Anomaly Detection via Domain-Aligned Memory Updates

Deploying video anomaly detection (VAD) in the real world is often constrained by the scarcity, privacy, and cost of collecting real abnormal footage. We propose PA-VAD, a novel pseudo-only framework that trains an anomaly detector without using any real abnormal videos, by pairing real normal videos with diffusion-synthesized pseudo-abnormal videos generated from a small set of real normal images. Beyond proposing a generation-driven training pipeline, we make a key empirical discovery: pseudo anomalies exhibit a characteristic spatiotemporal magnitude bias in feature space, which can dominate Multiple Instance Learning and degrade generalization if left unaddressed. To counter this pseudo-induced bias, we introduce the Domain-Aligned Regularized Module (DARM), which combines domain alignment with usage-aware memory updates to balance prototype coverage and stabilize optimization under biased pseudo supervision. Extensive experiments demonstrate that PA-VAD achieves 98.2% AUC on ShanghaiTech, 82.5% on UCF-Crime, and 95.1% on XD-Violence, and further improves generalization to unseen anomaly classes in open-set evaluations. Notably, PA-VAD surpasses the best real-abnormal WVAD baselines on ShanghaiTech and XD-Violence by +0.6% and +0.9%, respectively, and improves over the UVAD state of the art on UCF-Crime by +1.9% -showing that high-accuracy VAD is attainable without collecting real abnormal videos.

cs.CV↗
arXiv2023

Spatial profiles of collimated laser Compton-scattering $\gamma$-ray beams

The intensity and energy spatial distributions of collimated laser Compton scattering (LCS) $\gamma$-ray beams and of the associated bremsstrahlung beams have been investigated as functions of the electron beam energy, electron beam phase space distribution, laser optics conditions and laser polarization. We show that the beam halo is affected to different extents by variations in the above listed parameters. In the present work, we have used laser Compton scattering simulations performed with the \texttt{eliLaBr} code (https://github.com/dan-mihai-filipescu/eliLaBr) and real LCS and bremsstrahlung $\gamma$-ray beams produced at the NewSUBARU synchrotron radiation facility. A 500~$\mu$m MiniPIX X-ray camera was used as beamspot monitor in a wide $\gamma$-ray beam energy range between 1.73~MeV and 38.1~MeV.

physics.ins-det↗
arXiv2022

Spectral distribution and flux of $\gamma$-ray beams produced through Compton scattering of unsynchronized laser and electron beams

Intense, quasi-monochromatic, polarized $\gamma$-ray beams produced by Compton scattering of laser photons against relativistic electrons are used for fundamental studies and applications. Following a series of photoneutron cross section measurements in the Giant Dipole Resonance energy region performed at the NewSUBARU synchrotron radiation facility, we have developed the eliLaBr Monte Carlo simulation code for characterization of the scattered $\gamma$-ray photon beams. The code is implemented using Geant4 and is available on the GitHub repository (https://github.com/dan-mihai-filipescu/eliLaBr). Here we report the validation of the eliLaBr code on NewSUBARU LCS $\gamma$-ray beam flux and spectral distribution data and two applications performed with it for asymmetric transverse emittance profiles electron beams, characteristic for synchrotrons. The first application is based on a systematic investigation of transverse collimator offsets relative to the laser and electron beam axis. We show that the maximum energy of the LCS $\gamma$-ray beam is altered by vertical collimator offsets, where the edge shifts towards lower energies with the increase in the offset. Secondly, using the eliLaBr code, we investigate the effect of the laser polarization plane orientation on the properties of the LCS $\gamma$-ray beams produced with asymmetric emittance electron beams. We show that: 1. The use of vertically polarized lasers contributes to the preservation of the LCS $\gamma$-ray beam maximum energy edge by increasing the precision in the vertical collimator alignment. 2. Under identical conditions for the electron and laser beams phase-space distributions, the energy spectrum of the scattered LCS $\gamma$-ray beam changes with the laser beam polarization plane orientation: the use of vertically polarized laser beams slightly deteriorates the LCS $\gamma$-ray beam energy resolution.

physics.ins-det↗
arXiv2019

Model Predictive Control of Shallow Drowsiness: Improving Productivity of Office Workers

This paper proposes a methodology of model predictive control for alleviating shallow drowsiness of office workers and thus improving their productivity. The methodology is based on dynamically scheduling setting values for air conditioning and lighting to minimize drowsiness level of office workers on the basis of a prediction model that represents the relation between future drowsiness level and combination of indoor temperature and ambient illuminance. The prediction model can be identified by utilizing state-of-the-art drowsiness estimation method. The proposed methodology was evaluated in regard to a real routine task (performed by six subjects over five workdays), and the evaluation results demonstrate that the proposed methodology improved the processing speed of the task by 8.3% without degrading comfort of the workers.

eess.SP↗
arXiv2015

HARPO: beam characterization of a TPC for gamma-ray polarimetry and high angular-resolution astronomy in the MeV-GeV range

A time projection chamber (TPC) can be used to measure the polarization of gamma rays with excellent angular precision and sensitivity in the MeV-GeV energy range through the conversion of photons to e+e- pairs. The Hermetic ARgon POlarimeter (HARPO) prototype was built to demonstrate this concept. It was recently tested in the polarized photon beam at the NewSUBARU facility in Japan. We present this data-taking run, which demonstrated the excellent performance of the HARPO TPC.

astro-ph.IM↗