Zobrazeno 1 - 10
of 79
pro vyhledávání: '"Lin, Kun-Yu"'
We study the domain adaptation task for action recognition, namely domain adaptive action recognition, which aims to effectively transfer action recognition power from a label-sufficient source domain to a label-free target domain. Since actions are
Externí odkaz:
http://arxiv.org/abs/2407.10860
Learning generalizable visual dynamic representation across different embodied environments is crucial for real-world robotic manipulation. As the scale and diversity of robot demonstration data are limited, recent works have turned to large-scale pr
Externí odkaz:
http://arxiv.org/abs/2406.14235
Autor:
Lin, Kun-Yu, Ding, Henghui, Zhou, Jiaming, Tang, Yu-Ming, Peng, Yi-Xing, Zhao, Zhilin, Loy, Chen Change, Zheng, Wei-Shi
Building upon the impressive success of CLIP (Contrastive Language-Image Pretraining), recent pioneer works have proposed to adapt the powerful CLIP to video data, leading to efficient and effective video learners for open-vocabulary action recogniti
Externí odkaz:
http://arxiv.org/abs/2403.01560
Zero-shot action recognition (ZSAR) aims to learn an alignment model between videos and class descriptions of seen actions that is transferable to unseen actions. The text queries (class descriptions) used in existing ZSAR works, however, are often s
Externí odkaz:
http://arxiv.org/abs/2401.11654
Developing end-to-end action recognition models on long videos is fundamental and crucial for long-video action understanding. Due to the unaffordable cost of end-to-end training on the whole long videos, existing works generally train models on shor
Externí odkaz:
http://arxiv.org/abs/2311.17118
Video domain generalization aims to learn generalizable video classification models for unseen target domains by training in a source domain. A critical challenge of video domain generalization is to defend against the heavy reliance on domain-specif
Externí odkaz:
http://arxiv.org/abs/2310.17942
In this work, we focus on the task of procedure planning from instructional videos with text supervision, where a model aims to predict an action sequence to transform the initial visual state into the goal visual state. A critical challenge of this
Externí odkaz:
http://arxiv.org/abs/2308.08885
Autor:
Jiao, Jiayu, Tang, Yu-Ming, Lin, Kun-Yu, Gao, Yipeng, Ma, Jinhua, Wang, Yaowei, Zheng, Wei-Shi
Publikováno v:
IEEE Transaction on Multimedia, 2023
As a de facto solution, the vanilla Vision Transformers (ViTs) are encouraged to model long-range dependencies between arbitrary image patches while the global attended receptive field leads to quadratic computational cost. Another branch of Vision T
Externí odkaz:
http://arxiv.org/abs/2302.01791
Autor:
Du, Jia-Run, Feng, Jia-Chang, Lin, Kun-Yu, Hong, Fa-Ting, Wu, Xiao-Ming, Qi, Zhongang, Shan, Ying, Zheng, Wei-Shi
Weakly Supervised Temporal Action Localization (WSTAL) aims to localize and classify action instances in long untrimmed videos with only video-level category labels. Due to the lack of snippet-level supervision for indicating action boundaries, previ
Externí odkaz:
http://arxiv.org/abs/2206.11011
Deep neural networks for image classification only learn to map in-distribution inputs to their corresponding ground truth labels in training without differentiating out-of-distribution samples from in-distribution ones. This results from the assumpt
Externí odkaz:
http://arxiv.org/abs/2206.09385