This paper presents a domain generalization approach using meta-contrastive visual adaptation networks, enabling robust scene understanding without labeled target data across diverse visual environments and camera settings.

Lawrence Daniel Whitaker, Zhang Weitao, Rituja Anjali Desai, Anton Joseph Gruber, Hina Ayesha Khan, Haruto Kazuki Yamashita

This research introduces a hierarchical attention-based network for facial recognition tasks across domains with low-resolution images, achieving high accuracy through resolution-adaptive learning and feature refinement strategies.

Mitchell Andrew Hargrove, Liu Feiyan, Pooja Niranjan Deshpande, Maxime Claude Boucher, Chika Hoshiko Watanabe, Sofia Teresa Moreno

This study proposes lightweight pose estimation architectures optimized for edge computing, enabling real-time augmented reality experiences on low-power mobile and wearable devices without cloud dependency.

Graham Douglas Sinclair, Zhang Xueyan, Meera Shalini Patel, Lars Benjamin Eklund, Ayako Nanami Yoshida, Emilia Rose Cardoza

This work proposes a novel indoor navigation system that predicts scene graphs using semantic attention and object affordance cues, guiding autonomous agents through context-aware spatial planning in unknown interiors.

Daniel Thomas Falk, Chen Jiaqiang, Anika Ramesh Iyer, Jean-Pierre Laurent Beaulieu, Kojiro Masashi Tanaka, Helena Isabel Schmidt

This paper introduces a visual processing method that dynamically selects informative frames based on scene context to reduce energy consumption in UAV missions while preserving recognition accuracy in aerial tasks.

Philip Edward Monroe, Zhang Huiying, Radhika Suresh Vaidya, Tobias Alexander Lehmann, Hana Yukari Sasaki, Clara Giovanna Rossi

This study introduces a few-shot scene classification model tailored for disaster scenarios, combining vision transformers and contextual meta-learning to rapidly adapt to novel environments with minimal training samples.

Ronald James Cartwright, Xu Yifan, Priya Meenakshi Rao, Matteo Riccardo Lombardi, Akane Haruka Fujii, Juliette Noelle Becker

This research proposes a cross-device visual authentication framework leveraging multi-resolution feature correlation and identity-aware attention to enhance biometric matching reliability across varied camera types and resolutions.

Stephen Alan Whitmore, Liu Yuxiang, Sangeeta Nirmala Iyer, Thomas Olivier Lefevre, Emi Nanako Takahashi, Isabela Juliana Ortega