Zobrazeno 71 - 80
of 171
pro vyhledávání: '"Shou, Mike Zheng"'
Parameter-Efficient Transfer Learning (PETL) aims at efficiently adapting large models pre-trained on massive data to downstream tasks with limited task-specific data. In view of the practicality of PETL, previous works focus on tuning a small set of
Externí odkaz:
http://arxiv.org/abs/2303.07910
Learning object-centric representations from complex natural environments enables both humans and machines with reasoning abilities from low-level perceptual features. To capture compositional entities of the scene, we proposed cyclic walks between p
Externí odkaz:
http://arxiv.org/abs/2302.08023
Autor:
Li, Ming, Xu, Xiangyu, Fan, Hehe, Zhou, Pan, Liu, Jun, Liu, Jia-Wei, Li, Jiahe, Keppo, Jussi, Shou, Mike Zheng, Yan, Shuicheng
Existing methods of privacy-preserving action recognition (PPAR) mainly focus on frame-level (spatial) privacy removal through 2D CNNs. Unfortunately, they have two major drawbacks. First, they may compromise temporal dynamics in input videos, which
Externí odkaz:
http://arxiv.org/abs/2301.03046
Autor:
Wu, Jay Zhangjie, Ge, Yixiao, Wang, Xintao, Lei, Weixian, Gu, Yuchao, Shi, Yufei, Hsu, Wynne, Shan, Ying, Qie, Xiaohu, Shou, Mike Zheng
To replicate the success of text-to-image (T2I) generation, recent works employ large-scale video datasets to train a text-to-video (T2V) generator. Despite their promising results, such paradigm is computationally expensive. In this work, we propose
Externí odkaz:
http://arxiv.org/abs/2212.11565
Vision-Language Pre-Training (VLP) has shown promising capabilities to align image and text pairs, facilitating a broad variety of cross-modal learning tasks. However, we observe that VLP models often lack the visual grounding/localization capability
Externí odkaz:
http://arxiv.org/abs/2212.09737
To build Video Question Answering (VideoQA) systems capable of assisting humans in daily activities, seeking answers from long-form videos with diverse and complex events is a must. Existing multi-modal VQA models achieve promising performance on ima
Externí odkaz:
http://arxiv.org/abs/2212.09522
Autor:
Xu, Zhongcong, Zhang, Jianfeng, Liew, Jun Hao, Zhang, Wenqing, Bai, Song, Feng, Jiashi, Shou, Mike Zheng
Recent advances in generative adversarial networks (GANs) have demonstrated the capabilities of generating stunning photo-realistic portrait images. While some prior works have applied such image GANs to unconditional 2D portrait video generation and
Externí odkaz:
http://arxiv.org/abs/2212.06384
Vector-Quantized (VQ-based) generative models usually consist of two basic components, i.e., VQ tokenizers and generative transformers. Prior research focuses on improving the reconstruction fidelity of VQ tokenizers but rarely examines how the impro
Externí odkaz:
http://arxiv.org/abs/2212.03185
Autor:
Singh, Parantak, Li, You, Sikarwar, Ankur, Lei, Weixian, Gao, Daniel, Talbot, Morgan Bruce, Sun, Ying, Shou, Mike Zheng, Kreiman, Gabriel, Zhang, Mengmi
Curriculum design is a fundamental component of education. For example, when we learn mathematics at school, we build upon our knowledge of addition to learn multiplication. These and other concepts must be mastered before our first algebra lesson, w
Externí odkaz:
http://arxiv.org/abs/2211.15470
Autor:
Zhang, Binjie, Su, Shupeng, Ge, Yixiao, Xu, Xuyuan, Wang, Yexin, Yuan, Chun, Shou, Mike Zheng, Shan, Ying
The traditional model upgrading paradigm for retrieval requires recomputing all gallery embeddings before deploying the new model (dubbed as "backfilling"), which is quite expensive and time-consuming considering billions of instances in industrial a
Externí odkaz:
http://arxiv.org/abs/2210.06954