23 citations · 146 across the 33 of their papers we have counts for
14 papers · 1 filter
Learning Imbalanced Data with Vision Transformers
Zhengzhuo Xu, Ruikang Liu, Shuo Yang +2
The real-world data tends to be heavily imbalanced and severely skew the data-driven deep neural networks, which makes Long-Tailed Recognition (LTR) a massive challenging task. Exi…
Uni-Perceiver v2: A Generalist Model for Large-Scale Vision and Vision-Language Tasks
Hao Li, Jinguo Zhu, Xiaohu Jiang +8
Despite the remarkable success of foundation models, their task-specific fine-tuning paradigm makes them inconsistent with the goal of general perception modeling. The key to elimi…
High-fidelity Facial Avatar Reconstruction from Monocular Video with Generative Priors
Yunpeng Bai, Yanbo Fan, Xuan Wang +4
High-fidelity facial avatar reconstruction from a monocular video is a significant research problem in computer graphics and computer vision. Recently, Neural Radiance Field (NeRF)…
When Noisy Labels Meet Long Tail Dilemmas: A Representation Calibration Method
Manyi Zhang, Xuyang Zhao, Jun Yao +2
Real-world large-scale datasets are both noisily labeled and class-imbalanced. The issues seriously hurt the generalization of trained models. It is hence significant to address th…
Darwinian Model Upgrades: Model Evolving with Selective Compatibility
Binjie Zhang, Shupeng Su, Yixiao Ge +5
The traditional model upgrading paradigm for retrieval requires recomputing all gallery embeddings before deploying the new model (dubbed as "backfilling"), which is quite expensiv…
DeViT: Deformed Vision Transformers in Video Inpainting
Jiayin Cai, Changlin Li, Xin Tao +2
This paper proposes a novel video inpainting method. We make three main contributions: First, we extended previous Transformers with patch alignment by introducing Deformed Patch-b…
ViTKD: Practical Guidelines for ViT feature knowledge distillation
Zhendong Yang, Zhe Li, Ailing Zeng +3
Knowledge Distillation (KD) for Convolutional Neural Network (CNN) is extensively studied as a way to boost the performance of a small model. Recently, Vision Transformer (ViT) has…
Rethinking Knowledge Distillation via Cross-Entropy
Zhendong Yang, Zhe Li, Yuan Gong +4
Knowledge Distillation (KD) has developed extensively and boosted various tasks. The classical KD method adds the KD loss to the original cross-entropy (CE) loss. We try to decompo…
HyP Loss: Beyond Hypersphere Metric Space for Multi-label Image Retrieval
Chengyin Xu, Zenghao Chai, Zhengzhuo Xu +3
Image retrieval has become an increasingly appealing technique with broad multimedia application prospects, where deep hashing serves as the dominant branch towards low storage and…
Masked Generative Distillation
Zhendong Yang, Zhe Li, Mingqi Shao +3
Knowledge distillation has been applied to various tasks successfully. The current distillation algorithm usually improves students' performance by imitating the output of the teac…
Privacy-Preserving Model Upgrades with Bidirectional Compatible Training in Image Retrieval
Shupeng Su, Binjie Zhang, Yixiao Ge +4
The task of privacy-preserving model upgrades in image retrieval desires to reap the benefits of rapidly evolving new models without accessing the raw gallery images. A pioneering…
Towards Universal Backward-Compatible Representation Learning
Binjie Zhang, Yixiao Ge, Yantao Shen +6
Conventional model upgrades for visual search systems require offline refresh of gallery features by feeding gallery images into new models (dubbed as "backfill"), which is time-co…
REALY: Rethinking the Evaluation of 3D Face Reconstruction
Zenghao Chai, Haoxian Zhang, Jing Ren +5
The evaluation of 3D face reconstruction results typically relies on a rigid shape alignment between the estimated 3D model and the ground-truth scan. We observe that aligning two…
Hot-Refresh Model Upgrades with Regression-Alleviating Compatible Training in Image Retrieval
Binjie Zhang, Yixiao Ge, Yantao Shen +5
The task of hot-refresh model upgrades of image retrieval systems plays an essential role in the industry but has never been investigated in academia before. Conventional cold-refr…