Sorry, you need to enable JavaScript to visit this website.

Drowsiness is a major cause of vehicular and in
dustrial accidents, motivating the need for real-time monitoring
systems that are accurate, non-intrusive, and deployable on low
power platforms. While physiological signal–based approaches
(e.g., EEG/EOG) provide high reliability, their dependence on
sensor attachment makes them intrusive for everyday use.
Camera-based solutions are more practical but face two key
challenges.
First, state-of-the-art deep learning–based face recognition and

Reconstructing thin 3D structures is challenging due to their sparsity, scale variation, and complex geometry. Such structures arise in a wide range of domains, including medical imaging of vascular systems and industrial pipe systems. While recent neural methods perform well on dense surfaces, they often fail to recover fine thin geometries. We propose a reconstruction approach based on local depth projections, which provide an efficient and informative 2D representation of thin structures.

Driver fatigue remains a leading cause of road accidents, responsible for 24% of crashes. While yawning serves as an early behavioral indicator of fatigue, existing approaches face significant challenges due to the presence of systematic noise in video-annotated datasets arising from coarse temporal annotations. Training robust machine learning (ML) models requires rich supervisory labels that help learn salient features from the training data.

Reconstructing complete 3D geometry from monocular clinical endoscopic videos is challenging due to weak texture, repetitive tissue patterns, and severe illumination artifacts. Although emerging dense matching methods exhibit improved resilience to textureless regions, they often produce abundant spurious correspondences across non-overlapping views, which corrupts the correspondence graph and cause structure-from-motion (SfM) pipelines to fail.

Point clouds are a fundamental representation for 3D perception in robotics, autonomous systems, and augmented reality; however, despite progress in single-label classification, multi-label object recognition in point clouds—where multiple object categories inherently co-exist—remains underexplored.

White blood cell classification across heterogeneous image datasets remains challenging due to variations in staining protocols, scanning devices, resolution, and class distribution. In this work, we propose a two-stage framework for robust cross-dataset classification. YOLOv11 is first applied for white blood cell localization and extraction, followed by a Swin Transformer for fine-grained classification. Experimental results on multi-source datasets demonstrate that the proposed method outperforms ResNet50, VGG16-BN, and EfficientNet-B0.

Cooperative Perception (CP) supports autonomous driving by enabling connected autonomous vehicles (CAVs) to share sensor data, thereby increasing situational awareness and safety. However, training CP models requires diverse traffic scenarios, and real-world datasets are limited due to the high cost of multi-agent setups and the labor-intensive nature of labeling. Synthetic datasets offer a scalable alternative, but models trained exclusively on virtual data often face domain adaptation challenges.

Zero-shot anomaly detection aims to identify defects in unseen categories without target-specific training. Existing methods usually apply the same feature transformation to all samples, treating normal and anomalous data uniformly despite their fundamentally asymmetric distributions, compact normals versus diverse anomalies. We instead exploit this natural asymmetry by proposing AVA-DINO, an anomaly-aware vision-language adaptation framework with dual specialized branches for normal and anomalous patterns that adapt frozen DINOv3 visual features.

Active learning reduces annotation costs by prioritizing the most informative samples, however, methods that are using fixed batch sizes can be inefficient as they fail to adjust to variations in data and resource limitations. This challenge becomes even more critical in resource-constraint environments, such as IoT and edge devices, where both computational power and the availability of labeled data for training are limited. We propose Adaptive Batch Size using Reinforced Active Learning, a novel method that employs a dual-strategy design to dynamically optimize batch sizes.

Joint Embedding Predictive Architectures (JEPAs) learn visual
representations by predicting latent target embeddings from
contextual views, without reconstruction or contrastive negatives.
While effective, existing JEPA predictors are shallow feed-forward
networks applied once per training step, offering limited control
over multi-step latent dynamics and stability. We revisit this
design choice and introduce Learnable Iterated Function Systems
(LIFS), a structured predictor composed of a contractive mixture