… representations are essential for position-sensitive vision … representation through a
subnetwork that is formed by … high-resolution representation from the encoded low-resolution …

… Instead, we propose a universal representation learning face recognition framework, URFace,
that can deal with larger variations unseen in the given training data, without leveraging …

Computer vision has achieved remarkable success by (a) representing images as uniformly-arranged
pixel arrays and (b) convolving highly-localized features. However, convolutions …

… -to-token representation learning first proposed in our conference version with outlook
attention and presented a new model, Vision Outlooker (VOLO), for solving computer vision tasks. …

… visual representations … visionlanguage tasks [17] or vision recognition tasks [9, 32]. Our work
shares a similar format of visual representation with [17] while we focus on the area of vision-…

… the success of Mamba to vision, ie, building a generic vision backbone purely upon … Inspired
by ViT [14] and BERT [31], we also use class token to represent the whole patch sequence, …

… To address these issues, we propose a Semantic-Specific Graph Representation Learning
(… representations and 2) a semantic interaction module that correlates these representations …

… Feature representation and selection is a classic problem in computer vision and machine
learning [8]. Unlike feature representation in an image space, the feature representation of …

… label-efficient multimodal medical imaging representations by leveraging radiology reports.
… the learned representations for various downstream medical image recognition tasks with …

Much of vision-and-language research focuses on a small but diverse set of independent
tasks and supporting datasets often studied in isolation; however, the visually-grounded …