Filters








1,403 Hits in 4.9 sec

Training Object Detectors from Few Weakly-Labeled and Many Unlabeled Images [article]

Zhaohui Yang, Miaojing Shi, Chao Xu, Vittorio Ferrari, Yannis Avrithis
2021 arXiv   pre-print
In this work, we study the problem of training an object detector from one or few images with image-level labels and a larger set of completely unlabeled images.  ...  Our solution is to train a weakly-supervised student detector model from image-level pseudo-labels generated on the unlabeled set by a teacher classifier model, bootstrapped by region-level similarities  ...  Acknowledgement This work was partially supported by the National Natural Science Foundation of China (NSFC) under Grant No. 61828602 and 61876007.  ... 
arXiv:1912.00384v6 fatcat:tayfyoenv5gnzpd4btaprtn5dy

Semi-Weakly Supervised Object Detection by Sampling Pseudo Ground-Truth Boxes [article]

Akhil Meethal, Marco Pedersoli, Zhongwen Zhu, Francisco Perdigon Romero, Eric Granger
2022 arXiv   pre-print
These pseudo GT boxes are sampled from weakly-labeled images based on the categorical score of object proposals accumulated via a score propagation process.  ...  images with information in weakly-labeled images.  ...  Semi/Weakly-Supervised Object Detection: Semi-supervised object detectors rely on a small subset of labeled images and a large collection of unlabeled images to train a detector [29] .  ... 
arXiv:2204.00147v2 fatcat:qelz7qt4rrcfnbungejhserecy

Learning to Count Objects with Few Exemplar Annotations [article]

Jianfeng Wang, Rong Xiao, Yandong Guo, Lei Zhang
2019 arXiv   pre-print
Based on the observation that in many object counting problems the target objects are normally repeated and highly similar to each other, we are particularly interested in the setting when only a few exemplar  ...  On the CARPK dataset for parking lot car counting, we improved mAP@0.5 from 4.58% to 72.44% using only 5 training images each with 5 bounding boxes.  ...  To reduce the annotation effort, weakly supervised learning trains object detector only based on image-level labels.  ... 
arXiv:1905.07898v1 fatcat:b5fo5sty35hjhcajjex3kza6yy

Self-Learning for Player Localization in Sports Video [article]

Kenji Okuma and David G. Lowe and James J. Little
2013 arXiv   pre-print
We combine those image cues and discover additional labels automatically from unlabelled data.  ...  detecting sports players and improved tracking, when videos contain very few labelled images.  ...  The training procedure starts from initializing with a small set of labelled images and a large set of unlabelled images from sparsely labelled video data.  ... 
arXiv:1307.7198v1 fatcat:yyryezpb6relveca4xdfjkfhqq

BAOD: Budget-Aware Object Detection

Alejandro Pardo, Mengmeng Xu, Ali Thabet, Pablo Arbelaez, Bernard Ghanem
2021 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
We adopt a hybrid supervised learning framework to train the object detector from both these types of annotation.  ...  We investigate both optimization and learning-based methods to sample which images to annotate and what type of annotation (strongly or weakly supervised) to annotate them with.  ...  This detector works as a teacher that predicts objects in every image in the weakly labeled image set W t .  ... 
doi:10.1109/cvprw53098.2021.00137 fatcat:s3diafekevg6xnfs7oa72uu2fm

Improving Object Detection with Selective Self-supervised Self-training [article]

Yandong Li, Di Huang, Danfeng Qin, Liqiang Wang, Boqing Gong
2020 arXiv   pre-print
They fail to improve object detectors in their vanilla forms due to the domain gap between the Web images and curated datasets.  ...  On the other hand, we propose a novel learning method motivated by two parallel lines of work that explore unlabeled data for image classification: self-training and self-supervised learning.  ...  Following the procedure in [67] , we first train a teacher object detector f (I, θ * t ) from the labeled images, where θ * t stands for the network weights.  ... 
arXiv:2007.09162v2 fatcat:jubt7f4alje47ghb3c7jntpyfa

Detecting Human-Object Interaction with Mixed Supervision [article]

Suresh Kirthi Kumaraswamy
2020 arXiv   pre-print
Human object interaction (HOI) detection is an important task in image understanding and reasoning.  ...  It performs close to or even better than many fully-supervised methods by using a mixed amount of strong and weak annotations; furthermore, it outperforms representative state of the art weakly and fully-supervised  ...  [32] studied the problem for semantic image segmentation from a combination of few strongly labeled (pixel-level annotations) and many weakly labeled (image-level labels or bounding boxes) images.  ... 
arXiv:2011.04971v2 fatcat:wd2z5slrpnctbauhsvabfwjsam

BAOD: Budget-Aware Object Detection [article]

Alejandro Pardo, Mengmeng Xu, Ali Thabet, Pablo Arbelaez, Bernard Ghanem
2021 arXiv   pre-print
We adopt a hybrid supervised learning framework to train the object detector from both these types of annotation.  ...  We investigate both optimization and learning-based methods to sample which images to annotate and what type of annotation (strongly or weakly supervised) to annotate them with.  ...  This detector works as a teacher that predicts objects in every image in the weakly labeled image set W t .  ... 
arXiv:1904.05443v2 fatcat:2hwp6aor7fe3ldr6icvzmqcuki

Few-Example Object Detection with Model Communication

Xuanyi Dong, Liang Zheng, Fan Ma, Yi Yang, Deyu Meng
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
In this paper, we study object detection using a large pool of unlabeled images and only a few labeled images per category, named "few-example object detection".  ...  the state-of-the-art weakly-supervised approaches using a large number of image-level labels.  ...  The blue boxes in the top row contain the training images where the few labeled and the many unlabeled images are in the gray and yellow areas, Fig. 2 : 2 The working flow of our method when multi-modal  ... 
doi:10.1109/tpami.2018.2844853 pmid:29994192 fatcat:lhbnbbngnvbdrgmrw4oyjgdvna

Progressive Object Transfer Detection

Hao Chen, Yali Wang, Guoyou Wang, Xiang Bai, Yu Qiao
2019 IEEE Transactions on Image Processing  
In WSTD, we design a recurrent object labelling mechanism for learning to annotate weakly-labeled images.  ...  generalize this capacity by exploiting objects from wild images.  ...  Hence, it is preferable choice to use few fully-annotated images in LSTD and most weakly-annotated images in WSTD. Object Supervision from Warm-Up Detector.  ... 
doi:10.1109/tip.2019.2938680 fatcat:vcqeip2q3vc2vfzp7vfqshj3xq

Detecting Human-Object Interaction with Mixed Supervision

Suresh Kirthi Kumaraswamy, Miaojing Shi, Ewa Kijak
2021 2021 IEEE Winter Conference on Applications of Computer Vision (WACV)  
Human object interaction (HOI) detection is an important task in image understanding and reasoning.  ...  It is in a form of HOI triplet human, verb, object , requiring bounding boxes for human and object, and action between them for the task completion.  ...  [32] studied the problem for semantic image segmentation from a combination of few strongly labeled (pixel-level annotations) and many weakly labeled (image-level labels or bounding boxes) images.  ... 
doi:10.1109/wacv48630.2021.00127 fatcat:r4ri4f5rpfb5vgn3jicp2diqie

Beyond Weakly-supervised: Pseudo Ground Truths Mining for Missing Bounding-boxes Object Detection

Yongqiang Zhang, Mingli Ding, Yancheng Bai, Mengmeng Xu, Bernard Ghanem
2019 IEEE transactions on circuits and systems for video technology (Print)  
, and then combine the mined pseudo ground truths and the labeled annotations to train a fullysupervised object detector.  ...  Due to the shortcomings of the weakly-supervised and fully-supervised object detection (i.e. unsatisfactory performance and expensive annotations, respectively), leveraging partially labeled images in  ...  [50] transfers the tracked location of objects from the labeled videos to generated the boundingboxes for the weakly-labeled images.  ... 
doi:10.1109/tcsvt.2019.2898559 fatcat:hily5ccdvfettcr7wf2dsblxf4

A Weakly Supervised Method for Mud Detection in Ores Based on Deep Active Learning

Zhijian Huang, Fangmin Li, Xidao Luan, Zuowei Cai
2020 Mathematical Problems in Engineering  
Moreover, training a deep learning model needs a large amount of exactly labeled samples, which is expensive and time consuming.  ...  Aiming at the challenging problem, this paper proposed a novel weakly supervised method based on deep active learning (AL), named YOLO-AL.  ...  Acknowledgments is study was supported by the Scientific Research Fund of Hunan Provincial Education Department (nos. 18A376 and XJK17BXX010) and National Natural Science Foundation of China (no. 11701172  ... 
doi:10.1155/2020/3510313 fatcat:6klk3wmobbfr3ni3l4e6gtxuvm

Bayesian Joint Topic Modelling for Weakly Supervised Object Localisation [article]

Zhiyuan Shi, Timothy M. Hospedales, Tao Xiang
2017 arXiv   pre-print
supervision. (3) Our model can be learned with a mixture of weakly labelled and unlabelled data, allowing the large volume of unlabelled images on the Internet to be exploited for learning.  ...  We address the problem of localisation of objects as bounding boxes in images with weak labels.  ...  Refined by detector After the initial annotation of the weakly labelled images, a conventional strong object detector can be trained using these annotations as ground truth.  ... 
arXiv:1705.03372v1 fatcat:6gskbbogw5ejjfqsjvz5brw5wy

Bayesian Joint Topic Modelling for Weakly Supervised Object Localisation

Zhiyuan Shi, Timothy M. Hospedales, Tao Xiang
2013 2013 IEEE International Conference on Computer Vision  
supervision. (3) Our model can be learned with a mixture of weakly labelled and unlabelled data, allowing the large volume of unlabelled images on the Internet to be exploited for learning.  ...  We address the problem of localisation of objects as bounding boxes in images with weak labels.  ...  Refined by detector After the initial annotation of the weakly labelled images, a conventional strong object detector can be trained using these annotations as ground truth.  ... 
doi:10.1109/iccv.2013.371 dblp:conf/iccv/ShiHX13 fatcat:r2kkj7j72jfvzlcvpqyj5u2quy
« Previous Showing results 1 — 15 out of 1,403 results