Search

Your search keyword '"van de Weijer, Joost"' showing total 20 results
Start Over Author "van de Weijer, Joost" Publication Year Range Last 3 years Publisher ieee
20 results on '"van de Weijer, Joost"'

4. The First Visual Object Tracking Segmentation VOTS2023 Challenge Results

Author

Kristan, Matej, primary, Matas, Jiří, additional, Danelljan, Martin, additional, Felsberg, Michael, additional, Chang, Hyung Jin, additional, Čehovin Zajc, Luka, additional, Lukežič, Alan, additional, Drbohlav, Ondrej, additional, Zhang, Zhongqun, additional, Tran, Khanh-Tung, additional, Vu, Xuan-Son, additional, Björklund, Johanna, additional, Mayer, Christoph, additional, Zhang, Yushan, additional, Ke, Lei, additional, Zhao, Jie, additional, Fernández, Gustavo, additional, Al-Shakarji, Noor, additional, An, Dong, additional, Arens, Michael, additional, Becker, Stefan, additional, Bhat, Goutam, additional, Bullinger, Sebastian, additional, Chan, Antoni B., additional, Chang, Shijie, additional, Chen, Hanyuan, additional, Chen, Xin, additional, Chen, Yan, additional, Chen, Zhenyu, additional, Cheng, Yangming, additional, Cui, Yutao, additional, Deng, Chunyuan, additional, Dong, Jiahua, additional, Dunnhofer, Matteo, additional, Feng, Wei, additional, Fu, Jianlong, additional, Gao, Jie, additional, Han, Ruize, additional, Hao, Zeqi, additional, He, Jun-Yan, additional, He, Keji, additional, He, Zhenyu, additional, Hu, Xiantao, additional, Huang, Kaer, additional, Huang, Yuqing, additional, Jiang, Yi, additional, Kang, Ben, additional, Lan, Jin-Peng, additional, Lee, Hyungjun, additional, Li, Chenyang, additional, Li, Jiahao, additional, Li, Ning, additional, Li, Wangkai, additional, Li, Xiaodi, additional, Li, Xin, additional, Liu, Pengyu, additional, Liu, Yue, additional, Lu, Huchuan, additional, Luo, Bin, additional, Luo, Ping, additional, Ma, Yinchao, additional, Miao, Deshui, additional, Micheloni, Christian, additional, Palaniappan, Kannappan, additional, Park, Hancheol, additional, Paul, Matthieu, additional, Peng, HouWen, additional, Qian, Zekun, additional, Rahmon, Gani, additional, Scherer-Negenborn, Norbert, additional, Shao, Pengcheng, additional, Shin, Wooksu, additional, Kazemi, Elham Soltani, additional, Song, Tianhui, additional, Stiefelhagen, Rainer, additional, Sun, Rui, additional, Tang, Chuanming, additional, Tang, Zhangyong, additional, Toubal, Imad Eddine, additional, Valmadre, Jack, additional, van de Weijer, Joost, additional, Van Gool, Luc, additional, Vira, Jash, additional, Vujasinović, Stèphane, additional, Wan, Cheng, additional, Wan, Jia, additional, Wang, Dong, additional, Wang, Fei, additional, Wang, Feifan, additional, Wang, He, additional, Wang, Limin, additional, Wang, Song, additional, Wang, Yaowei, additional, Wang, Zhepeng, additional, Wu, Gangshan, additional, Wu, Jiannan, additional, Wu, Qiangqiang, additional, Wu, Xiaojun, additional, Xiao, Anqi, additional, Xie, Jinxia, additional, Xu, Chenlong, additional, Xu, Min, additional, Xu, Tianyang, additional, Xu, Yuanyou, additional, Yan, Bin, additional, Yang, Dawei, additional, Yang, Ming-Hsuan, additional, Yang, Tianyu, additional, Yang, Yi, additional, Yang, Zongxin, additional, Yin, Xuanwu, additional, Yu, Fisher, additional, Yu, Hongyuan, additional, Yu, Qianjin, additional, Yu, Weichen, additional, Yuan, YongSheng, additional, Yuan, Zehuan, additional, Zhang, Jianlin, additional, Zhang, Lu, additional, Zhang, Tianzhu, additional, Zhao, Guodongfang, additional, Zhao, Shaochuan, additional, Zheng, Yaozong, additional, Zhong, Bineng, additional, Zhu, Jiawen, additional, Zhu, Xuefeng, additional, Zhuang, Yueting, additional, Zong, ChengAo, additional, and Zuo, Kunlong, additional

Published
2023
Full Text
View/download PDF

12. Towards Exemplar-Free Continual Learning in Vision Transformers: an Account of Attention, Functional and Weight Regularization

Author

Pelosin, Francesco, Jha, Saurav, Torsello, Andrea, Raducanu, Bogdan, and van de Weijer, Joost

Subjects
FOS: Computer and information sciences, Computer Science - Machine Learning, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition, Machine Learning (cs.LG)
Abstract

In this paper, we investigate the continual learning of Vision Transformers (ViT) for the challenging exemplar-free scenario, with special focus on how to efficiently distill the knowledge of its crucial self-attention mechanism (SAM). Our work takes an initial step towards a surgical investigation of SAM for designing coherent continual learning methods in ViTs. We first carry out an evaluation of established continual learning regularization techniques. We then examine the effect of regularization when applied to two key enablers of SAM: (a) the contextualized embedding layers, for their ability to capture well-scaled representations with respect to the values, and (b) the prescaled attention maps, for carrying value-independent global contextual information. We depict the perks of each distilling strategy on two image recognition benchmarks (CIFAR100 and ImageNet-32) -- while (a) leads to a better overall accuracy, (b) helps enhance the rigidity by maintaining competitive performances. Furthermore, we identify the limitation imposed by the symmetric nature of regularization losses. To alleviate this, we propose an asymmetric variant and apply it to the pooled output distillation (POD) loss adapted for ViTs. Our experiments confirm that introducing asymmetry to POD boosts its plasticity while retaining stability across (a) and (b). Moreover, we acknowledge low forgetting measures for all the compared methods, indicating that ViTs might be naturally inclined continual learner, Accepted at Continual Learning Workshop (CVPR 2022)

Published
2022

13. Continually Learning Self-Supervised Representations with Projected Functional Regularization

Author

Gomez-Villa, Alex, Twardowski, Bartlomiej, Yu, Lu, Bagdanov, Andrew D., and van de Weijer, Joost

Subjects
FOS: Computer and information sciences, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition
Abstract

Recent self-supervised learning methods are able to learn high-quality image representations and are closing the gap with supervised approaches. However, these methods are unable to acquire new knowledge incrementally -- they are, in fact, mostly used only as a pre-training phase over IID data. In this work we investigate self-supervised methods in continual learning regimes without any replay mechanism. We show that naive functional regularization, also known as feature distillation, leads to lower plasticity and limits continual learning performance. Instead, we propose Projected Functional Regularization in which a separate temporal projection network ensures that the newly learned feature space preserves information of the previous one, while at the same time allowing for the learning of new features. This prevents forgetting while maintaining the plasticity of the learner. Comparison with other incremental learning approaches applied to self-supervision demonstrates that our method obtains competitive performance in different scenarios and on multiple datasets., Comment: Accepted at Workshop on Continual Learning in Computer Vision (CVPR 2022)

Published
2022

18. TransferI2I: Transfer Learning for Image-to-Image Translation from Small Datasets

Author

Wang, Yaxing, Mantecon, Hector Laria, van de Weijer, Joost, Lopez-Fuentes, Laura, and Raducanu, Bogdan

Subjects
FOS: Computer and information sciences, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition
Abstract

Image-to-image (I2I) translation has matured in recent years and is able to generate high-quality realistic images. However, despite current success, it still faces important challenges when applied to small domains. Existing methods use transfer learning for I2I translation, but they still require the learning of millions of parameters from scratch. This drawback severely limits its application on small domains. In this paper, we propose a new transfer learning for I2I translation (TransferI2I). We decouple our learning process into the image generation step and the I2I translation step. In the first step we propose two novel techniques: source-target initialization and self-initialization of the adaptor layer. The former finetunes the pretrained generative model (e.g., StyleGAN) on source and target data. The latter allows to initialize all non-pretrained network parameters without the need of any data. These techniques provide a better initialization for the I2I translation step. In addition, we introduce an auxiliary GAN that further facilitates the training of deep I2I systems even from small datasets. In extensive experiments on three datasets, (Animal faces, Birds, and Foods), we show that we outperform existing methods and that mFID improves on several datasets with over 25 points., Technical report

Published
2021

Catalog

Books, media, physical & digital resources
See catalog results