Filters








115,274 Hits in 4.2 sec

A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark [article]

Xiaohua Zhai, Joan Puigcerver, Alexander Kolesnikov, Pierre Ruyssen, Carlos Riquelme, Mario Lucic, Josip Djolonga, Andre Susano Pinto, Maxim Neumann, Alexey Dosovitskiy, Lucas Beyer, Olivier Bachem, Michael Tschannen, Marcin Michalski (+3 others)
2020 arXiv   pre-print
We present the Visual Task Adaptation Benchmark (VTAB), which defines good representations as those that adapt to diverse, unseen tasks with few examples.  ...  Yet, the absence of a unified evaluation for general visual representations hinders progress.  ...  Supplementary Material: A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark A.  ... 
arXiv:1910.04867v2 fatcat:wt57k37z3ncb7iugni26qd7roi

Guest Editorial Introduction to the Special Section on Video and Language

Tao Mei, Jason J. Corso, Gunhee Kim, Jiebo Luo, Chunhua Shen, Hanwang Zhang
2022 IEEE transactions on circuits and systems for video technology (Print)  
This task-adaptive attention module integrates the input image feature vectors with task-adaptive vectors, targeting for learning useful non-visual clues.  ...  Extensive experiments on the image captioning benchmark (MSCOCO) demonstrate the effectiveness of proposed task-adaptive attention module.  ...  Tao Mei (Fellow, IEEE) is currently the Vice President of JD.com and the Deputy Managing Director of JD Explore Academy.  ... 
doi:10.1109/tcsvt.2021.3137430 fatcat:ksel3hruujgwfpwalj4u5ebebu

MAGMA – Multimodal Augmentation of Generative Models through Adapter-based Finetuning [article]

Constantin Eichenberg, Sidney Black, Samuel Weinbach, Letitia Parcalabescu, Anette Frank
2021 arXiv   pre-print
MAGMA outperforms Frozen on open-ended generative tasks, achieving state of the art results on the OKVQA benchmark and competitive results on a range of other popular VL benchmarks, while pretraining on  ...  We present MAGMA - a simple method for augmenting generative language models with additional modalities using adapter-based finetuning.  ...  The adapter variant where an equal number of parameters are allocated to the attention and the feed forward adapters excels at the GQA benchmark, a question answering benchmark built around scene graphs  ... 
arXiv:2112.05253v1 fatcat:2pjd6mfwrvbsxkzfdxbflwiy54

Adaptive Objectness for Object Tracking

Pengpeng Liang, Yu Pang, Chunyuan Liao, Xue Mei, Haibin Ling
2016 IEEE Signal Processing Letters  
More specifically, we use the newly proposed BING objectness as the base, and then train an object-adaptive objectness for each tracking task.  ...  Thus motivated, in this paper we propose to adapt objectness for visual object tracking.  ...  CVPR2013 Visual Tracking Benchmark The CVPR2013 Visual Tracking Benchmark [37] includes 50 fully annotated sequences.  ... 
doi:10.1109/lsp.2016.2556706 fatcat:xkcsg4ce2jajxeaszbfmmqzmwu

Syn2Real: A New Benchmark forSynthetic-to-Real Visual Domain Adaptation [article]

Xingchao Peng, Ben Usman, Kuniaki Saito, Neela Kaushik, Judy Hoffman, Kate Saenko
2018 arXiv   pre-print
Our evaluation of multiple state-of-the-art methods reveals a large gap in adaptation performance between the easier closed-set classification task and the more difficult open-set and detection tasks.  ...  Unfortunately, current benchmarks for this problem are limited in size and task diversity.  ...  Our hope is that the Syn2Real benchmark will provide a useful tool for the visual domain adaptation and transfer learning community.  ... 
arXiv:1806.09755v1 fatcat:ekeh4gaj3vcdrezkgs6fafonme

Unsupervised Visual Attention and Invariance for Reinforcement Learning [article]

Xudong Wang, Long Lian, Stella X. Yu
2021 arXiv   pre-print
Control (our DrawerWorld Manipulation) benchmarks.  ...  Our Visual Attention and Invariance (VAI) method significantly outperforms the state-of-the-art on visual domain generalization, gaining 15 to 49% (61 to 229%) more cumulative rewards per episode on DeepMind  ...  The Grid task is the training task for the adapter.  ... 
arXiv:2104.02921v2 fatcat:ae5oftwoe5hozm2fqbrukox77e

MEmoBERT: Pre-training Model with Prompt-based Learning for Multimodal Emotion Recognition [article]

Jinming Zhao, Ruichen Li, Qin Jin, Xinchao Wang, Haizhou Li
2021 arXiv   pre-print
the downstream task closer to the pre-training.  ...  Furthermore, unlike the conventional "pre-train, finetune" paradigm, we propose a prompt-based method that reformulates the downstream emotion classification task as a masked text prediction one, bringing  ...  Once the model is well pre-trained, we adopt the prompt-based learning method to adapt it to downstream tasks.  ... 
arXiv:2111.00865v1 fatcat:pzlft4ufwzgplb6gceehaz4sxm

XDBERT: Distilling Visual Information to BERT from Cross-Modal Systems to Improve Language Understanding [article]

Chan-Jan Hsu, Hung-yi Lee, Yu Tsao
2022 arXiv   pre-print
Our framework is inspired by cross-modal encoders' success in visual-language tasks while we alter the learning objective to cater to the language-heavy characteristics of NLU.  ...  Transformer-based models are widely used in natural language understanding (NLU) tasks, and multimodal transformers have been effective in visual-language tasks.  ...  We thank the National Center for High-performance Computing (NCHC) of National Applied Research Laboratories (NARLabs) in Taiwan for providing computational and storage resources.  ... 
arXiv:2204.07316v3 fatcat:vyc5635tozegplcz4b5bxu4lny

Revisiting Image Aesthetic Assessment via Self-Supervised Feature Learning [article]

Kekai Sheng, Weiming Dong, Menglei Chai, Guohui Wang, Peng Zhou, Feiyue Huang, Bao-Gang Hu, Rongrong Ji, Chongyang Ma
2019 arXiv   pre-print
The features from our pretext tasks are then adapted for a one-layer linear classifier to evaluate the performance in terms of binary aesthetic classification.  ...  Visual aesthetic assessment has been an active research field for decades.  ...  The following adaptation stage shares the same settings except that the learning-rate starts from 0.01. Benchmarks for Aesthetic Assessment Aesthetic Visual Analysis (AVA).  ... 
arXiv:1911.11419v1 fatcat:626ud7lr75hmrhomqonaok2yua

Revisiting Image Aesthetic Assessment via Self-Supervised Feature Learning

Kekai Sheng, Weiming Dong, Menglei Chai, Guohui Wang, Peng Zhou, Feiyue Huang, Bao-Gang Hu, Rongrong Ji, Chongyang Ma
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
The features from our pretext tasks are then adapted for a one-layer linear classifier to evaluate the performance in terms of binary aesthetic classification.  ...  Visual aesthetic assessment has been an active research field for decades.  ...  The following adaptation stage shares the same settings except that the learning-rate starts from 0.01. Benchmarks for Aesthetic Assessment Aesthetic Visual Analysis (AVA).  ... 
doi:10.1609/aaai.v34i04.6026 fatcat:pmogku62tjfvbk5t6cipijjpzy

Robust Continuous System Integration for Critical Deep-Sea Robot Operations Using Knowledge-Enabled Simulation in the Loop [article]

Christian A. Mueller, Tobias Doernbach, Arturo Gomez Chavez, Daniel Koehntopp, Andreas Birk
2018 arXiv   pre-print
The conducted evaluation shows the benefit of the proposed work in tasks related to perception and self-localization under changing spatial and environmental conditions.  ...  As a consequence, measures within the development stage have to be implemented to extensively evaluate and benchmark system components ranging from data acquisition, perception and localization to control  ...  For visual survey and navigation tasks this error is minimal.  ... 
arXiv:1803.02127v2 fatcat:ctbsgnfx6jd6jdrfeuxpe333sy

Towards More Flexible and Accurate Object Tracking with Natural Language: Algorithms and Benchmark [article]

Xiao Wang, Xiujun Shu, Zhipeng Zhang, Bo Jiang, Yaowei Wang, Yonghong Tian, Feng Wu
2021 arXiv   pre-print
We also introduce two new challenges into TNL2K for the object tracking task, i.e., adversarial samples and modality switch.  ...  A strong baseline method based on an adaptive local-global-search scheme is proposed for future works to compare.  ...  Therefore, this benchmark can be only used for the task of tracking by joint language and bbox.  ... 
arXiv:2103.16746v1 fatcat:hxmyyxiaenhyjhyitwiazjkyea

Adversarial Networks with Circular Attention Mechanism for Fine-Grained Domain Adaptation

Ningyu He, Jie Zhu
2021 IEEE Access  
Fine-grained Image Analysis (FGIA) as a branch of the image analysis tasks has received more and more attention in recent years.  ...  An effective solution is to apply the domain adaptation (DA) method to transfer knowledge from existing fine-grained image datasets to massive unlabeled data.  ...  Figure 4 , 5 and 6 show example images for fine-grained domain adaptation task in the three benchmarks. 3) BENCHMARK III: BIRDS-31 Benchmark III (Birds-31) can also be split into two domains: NABirds  ... 
doi:10.1109/access.2021.3118786 fatcat:kgdhxxzdqzdilbhmbgft3ixvvy

Toward measuring visualization insight

C. North
2006 IEEE Computer Graphics and Applications  
Insight: The capacity to discern the true nature of a situation; The act or outcome of grasping the inward or hidden nature of things or of perceiving in an intuitive manner. -Merriam-Webster  ...  An example benchmark task might ask users to find the maximum value in the data set. The two primary independent variables are the visualization design alternatives and benchmark tasks.  ...  The choice of tasks and the phrasing of task questions can introduce bias toward one of the visualization designs. Benchmark tasks lack completeness.  ... 
doi:10.1109/mcg.2006.70 pmid:16711210 fatcat:5vr3dhsckfbuflcgjgmntrkije

A Broad Study of Pre-training for Domain Generalization and Adaptation [article]

Donghyun Kim, Kaihong Wang, Stan Sclaroff, Kate Saenko
2022 arXiv   pre-print
Thus, existing works pay little attention to the effects of pre-training on domain transfer tasks.  ...  We observe that simply using a state-of-the-art backbone outperforms existing state-of-the-art domain adaptation baselines and set new baselines on Office-Home and DomainNet improving by 10.7\% and 5.5  ...  We use features directly obtained from each pre-training without fine-tuning on domain adaptation benchmarks. Feature Visualization.  ... 
arXiv:2203.11819v2 fatcat:7hdtwqiqd5ht3p52ab4stzphnm
« Previous Showing results 1 — 15 out of 115,274 results