Filters








32,277 Hits in 8.0 sec

Incremental Multi-Target Domain Adaptation for Object Detection with Efficient Domain Transfer [article]

Le Thanh Nguyen-Meidine, Madhu Kiran, Marco Pedersoli, Jose Dolz, Louis-Antoine Blais-Morin, Eric Granger
2022 arXiv   pre-print
While the problem of single-target domain adaptation (STDA) for object detection has recently received much attention, multi-target domain adaptation (MTDA) remains largely unexplored, despite its practical  ...  In this paper, we introduce an efficient approach for incremental learning that generalizes well to multiple target domains.  ...  While this can significantly improve the results of STDA for object detection, it Multi-Target Domain Adaptation: Current approaches for MTDA mainly focus on the classification task.  ... 
arXiv:2104.06476v4 fatcat:fkmpfdkpxnfqpouhad7nkbzj34

Exploring multi-modality structure for cross domain adaptation in video concept annotation

Shaoxi Xu, Sheng Tang, Yongdong Zhang, Jintao Li, Yan-Tao Zheng
2012 Neurocomputing  
To our best knowledge, it is the first time to introduce multi-modality transfer into the field of domain adaptive video concept detection and annotation.  ...  classifiers in the source domains to assist multi-graph optimization (a graph-based semi-supervised learning method) in the target domain for video concept annotation.  ...  We compare our proposed MMT-MGO with the OMG-SSL which utilizes multi-modalities for annotation without domain adaptation and other three domain adaptive concept detection methods without multi-modality  ... 
doi:10.1016/j.neucom.2011.05.041 fatcat:dog64kb5qvafxfxzgs6f2y3weq

Context-Transformer: Tackling Object Confusion for Few-Shot Detection [article]

Ze Yang, Yu Qiao ShenZhen Key Lab of Computer Vision and Pattern Recognition, SIAT-SenseTime Joint Lab, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, SIAT Branch, Shenzhen Institute of Artificial Intelligence, Robotics for Society)
2020 arXiv   pre-print
However, such transferred detector often fails to recognize new objects in the target domain, due to low data diversity of training samples.  ...  Few-shot object detection is a challenging but realistic scenario, where only a few annotated training images are available for training detectors.  ...  Hence, they may lack the adaptation capacity for few-shot object detection. Source Detection Transfer To begin with, we formulate few-shot object detection in a practical transfer learning setting.  ... 
arXiv:2003.07304v1 fatcat:ndmy6jxsa5amthovf2dhyx2lai

Context-Transformer: Tackling Object Confusion for Few-Shot Detection

Ze Yang, Yali Wang, Xianyu Chen, Jianzhuang Liu, Yu Qiao
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
However, such transferred detector often fails to recognize new objects in the target domain, due to low data diversity of training samples.  ...  Few-shot object detection is a challenging but realistic scenario, where only a few annotated training images are available for training detectors.  ...  However, it relies on multi-model fusion with a complex training procedure, which may reduce the efficiency of model deployment for a new few-shot detection task.  ... 
doi:10.1609/aaai.v34i07.6957 fatcat:zx33qcdnxfcexpktk5bdp64nqi

Online Domain Adaptation for Multi-Object Tracking [article]

Adrien Gaidon, Eleonora Vig
2015 arXiv   pre-print
This work is the first to investigate the problem of on-line domain adaptation of object detectors for causal multi-object tracking (MOT).  ...  Automatically detecting, labeling, and tracking objects in videos depends first and foremost on accurate category-level object detectors.  ...  Our method is labeled ODAMOT, for "Online Domain Adaption for Multi-Object Tracking" (cf.  ... 
arXiv:1508.00776v1 fatcat:asjfnza2gbf6di2jncl6h77dme

Multi-Task Incremental Learning for Object Detection [article]

Xialei Liu, Hao Yang, Avinash Ravichandran, Rahul Bhotika, Stefano Soatto
2020 arXiv   pre-print
In this work, we propose three incremental learning scenarios across various domains and categories for object detection.  ...  Training an object detector incrementally across various domains has rarely been explored.  ...  Preliminaries In this section, we introduce the notation for object detection in multi-task incremental learning and present the proposed approach in Sec. 4.  ... 
arXiv:2002.05347v3 fatcat:erbi7aszjbgepmbofabcvahlam

SHIFT: A Synthetic Driving Dataset for Continuous Multi-Task Domain Adaptation [article]

Tao Sun, Mattia Segu, Janis Postels, Yuxuan Wang, Luc Van Gool, Bernt Schiele, Federico Tombari, Fisher Yu
2022 arXiv   pre-print
In this paper, we introduce the largest multi-task synthetic dataset for autonomous driving, SHIFT.  ...  Featuring a comprehensive sensor suite and annotations for several mainstream perception tasks, SHIFT allows investigating the degradation of a perception system performance at increasing levels of domain  ...  We evaluate four adaptation strategies: targeted domain adaptation (Targeted DA), untargeted domain adaptation (Untargeted DA), incremental domain adaptation (Incremental DA) and continuous test-time adaptation  ... 
arXiv:2206.08367v1 fatcat:qdajmi3tpzcsrcx7x2oo3ofsfy

On Generalizing Detection Models for Unconstrained Environments [article]

Prajjwal Bhargava
2019 arXiv   pre-print
We address the problem of incremental learning in object detection on the India Driving Dataset (IDD).  ...  Results show the effectiveness of our domain adaptive approach in the case of domain shifts in environments.  ...  Datasets IDD: We use IDD for target adaptation tasks. It provides data for object detection in two resolutions.  ... 
arXiv:1909.13080v1 fatcat:iz7r5vtamvd3flua67w67lffge

Domain Adaptation for Visual Applications: A Comprehensive Survey [article]

Gabriela Csurka
2017 arXiv   pre-print
The aim of this paper is to give an overview of domain adaptation and transfer learning with a specific view on visual applications.  ...  After a general motivation, we first position domain adaptation in the larger transfer learning problem.  ...  [199] uses the TCA [14] to adapt image level HOG representation between source and target domains for object detection.  ... 
arXiv:1702.05374v2 fatcat:5va4oz4evjfhxgxddflpbb6pxi

Zero-Annotation Object Detection with Web Knowledge Transfer [chapter]

Qingyi Tao, Hao Yang, Jianfei Cai
2018 Lecture Notes in Computer Science  
First of all, we propose an instance-level adversarial domain adaptation network with attention on foreground objects to transfer the object appearances from web domain to target domain.  ...  In order to facilitate effective knowledge transfer from web images, we introduce a multi-instance multi-label domain adaption learning framework with two key innovations.  ...  For example, the "cow" from web domain will be confused with the "sheep" from target domain through the domain adaptation.  ... 
doi:10.1007/978-3-030-01252-6_23 fatcat:eynxmqzi4fcgxe7tya5yobxb3m

Zero-Annotation Object Detection with Web Knowledge Transfer [article]

Qingyi Tao, Hao Yang, Jianfei Cai
2018 arXiv   pre-print
First of all, we propose an instance-level adversarial domain adaptation network with attention on foreground objects to transfer the object appearances from web domain to target domain.  ...  In order to facilitate effective knowledge transfer from web images, we introduce a multi-instance multi-label domain adaption learning framework with two key innovations.  ...  For example, the "cow" from web domain will be confused with the "sheep" from target domain through the domain adaptation.  ... 
arXiv:1711.05954v2 fatcat:h6t5z4ropfhmpktt3mry23vfqq

Self-Learning Scene-Specific Pedestrian Detectors Using a Progressive Latent Model

Qixiang Ye, Tianliang Zhang, Wei Ke, Qiang Qiu, Jie Chen, Guillermo Sapiro, Baochang Zhang
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
With the difference of convex (DC) objective functions, PLM can be efficiently optimized with a concaveconvex programming and thus guaranteeing the stability of self-learning.  ...  Extensive experiments demonstrate that even without annotation the proposed self-learning approach outperforms weakly supervised learning approaches, while achieving comparable performance with transfer  ...  However, transfer learning is challenged when the object appearance in the target domains has significant differences with that in the source domains; while semisupervised models might drift away from  ... 
doi:10.1109/cvpr.2017.222 dblp:conf/cvpr/YeZKQCSZ17 fatcat:4b6uc7qvfvhkhj5gljgim5a7ai

Sim-to-Real Transfer in Deep Reinforcement Learning for Robotics: a Survey [article]

Wenshuai Zhao, Jorge Peña Queralta, Tomi Westerlund
2020 arXiv   pre-print
Multiple research efforts are therefore now being directed towards closing this sim-to-real gap and accomplish more efficient policy transfer.  ...  In this survey paper, we cover the fundamental background behind sim-to-real transfer in deep reinforcement learning and overview the main methods being utilized at the moment: domain randomization, domain  ...  ACKNOWLEDGEMENTS This work was supported by the Academy of Finland's AutoSOS project with grant number 328755.  ... 
arXiv:2009.13303v1 fatcat:7xjickbrh5avlohasquqyxlhrq

An Efficient Domain-Incremental Learning Approach to Drive in All Weather Conditions [article]

M. Jehanzeb Mirza, Marc Masana, Horst Possegger, Horst Bischof
2022 arXiv   pre-print
We show the efficacy of our approach by testing it for object detection in a challenging domain-incremental autonomous driving scenario where we encounter different adverse weather conditions, such as  ...  When adapting these models for changed environments, such as different weather conditions, they are prone to forgetting previously learned information.  ...  Acknowledgments This work was partially funded by the Christian Doppler Laboratory for Embedded Machine Learning and the Austrian Research Promotion Agency (FFG) under the project High-Scene (884306).  ... 
arXiv:2204.08817v2 fatcat:t7yk2kxilfcfbpzdrrreoa6db4

DSDANet: Deep Siamese Domain Adaptation Convolutional Neural Network for Cross-domain Change Detection [article]

Hongruixuan Chen and Chen Wu and Bo Du and Liangpei Zhang
2020 arXiv   pre-print
How to learn a transferable CD model in the data set with enough labeled data (original domain) but can well detect changes in another data set without labeled data (target domain)?  ...  This is defined as the cross-domain change detection problem. In this paper, we propose a novel deep siamese domain adaptation convolutional neural network (DSDANet) architecture for cross-domain CD.  ...  CD. 2) In DSDANet, for enhancing the transferability of differ- ence feature to cope with some situations of severe do- main distribution discrepancy, a multi-kernel MMD with multiple-layer domain adaptation  ... 
arXiv:2006.09225v1 fatcat:mntn5yzdfbhz3jiemxemhodgkq
« Previous Showing results 1 — 15 out of 32,277 results