PooDLe：从自然视频中进行池化与密集自监督学习
PooDLe: Pooled and dense self-supervised learning from naturalistic videos
AI总结：
提出PooDLe方法，结合池化不变性与密集光流等变目标，在多个特征尺度上统一训练，从自然视频中学习图像表示，并在BDD100K和Walking Tours数据集上验证了空间与语义理解能力。
摘要：
自监督学习在从单主体、标志性图像中学习方面取得了显著进展。然而，关于使用最小化筛选的自然视频数据，仍存在未解答的问题，这些数据包含具有许多独立对象的密集场景、不平衡的类别分布以及不同大小的对象。在本文中，我们提出了PooDLe，一种自监督学习方法，它将基于不变性的池化表示目标与强制光流扭曲等变性的密集自监督目标相结合。我们的结果表明，在多个特征尺度上应用统一目标对于从自然视频中学习有效的图像表示至关重要。我们通过在BDD100K驾驶视频数据集和Walking Tours第一人称视频数据集上的实验验证了我们的方法，展示了其通过密集目标捕获空间理解以及通过池化表示目标捕获语义理解的能力。
英文摘要：
Self-supervised learning has driven significant progress in learning from single-subject, iconic images. However, there are still unanswered questions about the use of minimally-curated, naturalistic video data, which contain dense scenes with many independent objects, imbalanced class distributions, and varying object sizes. In this paper, we propose PooDLe, a self-supervised learning method that combines an invariance-based objective on pooled representations with a dense SSL objective that enforces equivariance to optical flow warping. Our results show that a unified objective applied at multiple feature scales is essential for learning effective image representations from naturalistic videos. We validate our method with experiments on the BDD100K driving video dataset and the Walking Tours first-person video dataset, demonstrating its ability to capture spatial understanding from a dense objective and semantic understanding via a pooled representation objective.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
