📊 Статистика дайджестов

Всего дайджестов: 34022 Добавлено сегодня: 82

Последнее обновление: сегодня
Авторы:

Qi Jiang, Xiaolong Qian, Yao Gao, Lei Sun, Kailun Yang, Zhonghua Yi, Wenyong Li, Ming-Hsuan Yang, Luc Van Gool, Kaiwei Wang

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Emerging deep-learning-based lens library pre-training (LensLib-PT) pipeline offers a new avenue for blind lens aberration correction by training a universal neural network, demonstrating strong capability in handling diverse unknown optical degradations. This work proposes the OmniLens++ framework, which resolves two challenges that hinder the generalization ability of existing pipelines: the difficulty of scaling data and the absence of prior guidance characterizing optical degradation. To imp...
ID: 2511.17126v3 eess.IV, cs.CV, cs.LG, physics.optics
Авторы:

Qi Jiang, Xiaolong Qian, Yao Gao, Lei Sun, Kailun Yang, Zhonghua Yi, Wenyong Li, Ming-Hsuan Yang, Luc Van Gool, Kaiwei Wang

Саммари на русском не найдено
Доступные поля: ['id', 'arxiv_id', 'title', 'authors', 'abstract', 'summary_ru', 'categories', 'published_date', 'created_at']
Annotation:
Emerging deep-learning-based lens library pre-training (LensLib-PT) pipeline offers a new avenue for blind lens aberration correction by training a universal neural network, demonstrating strong capability in handling diverse unknown optical degradations. This work proposes the OmniLens++ framework, which resolves two challenges that hinder the generalization ability of existing pipelines: the difficulty of scaling data and the absence of prior guidance characterizing optical degradation. To imp...
ID: 2511.17126v1 eess.IV, cs.CV, cs.LG, physics.optics