Filters








1,430 Hits in 4.1 sec

Bounding boxes for weakly supervised segmentation: Global constraints get close to full supervision [article]

Hoel Kervadec, Jose Dolz, Shanshan Wang, Eric Granger, Ismail Ben Ayed
2020 arXiv   pre-print
We propose a novel weakly supervised learning segmentation based on several global constraints derived from box annotations.  ...  Furthermore, we integrate our deep tightness prior with a global background emptiness constraint, guiding training with information outside the bounding box.  ...  Acknowledgments This work is supported by the National Science and Engineering Research Council of Canada (NSERC), via its Discovery Grant program. We also thank NVIDIA for the GPU donation.  ... 
arXiv:2004.06816v1 fatcat:tqoiditdbneotefr5waolf7qba

Bounding Box Tightness Prior for Weakly Supervised Image Segmentation [chapter]

Juan Wang, Bin Xia
2021 Lecture Notes in Computer Science  
This paper presents a weakly supervised image segmentation method that adopts tight bounding box annotations.  ...  It proposes generalized multiple instance learning (MIL) and smooth maximum approximation to integrate the bounding box tightness prior into the deep neural network in an end-to-end manner.  ...  MIL baseline MIL definition and bounding box tightness prior Multiple instance learning (MIL) is a type of supervised learning.  ... 
doi:10.1007/978-3-030-87196-3_49 fatcat:uyhobw3bfzaybf2ee3z3wn5jte

Polar Transformation Based Multiple Instance Learning Assisting Weakly Supervised Image Segmentation With Loose Bounding Box Annotations [article]

Juan Wang, Bin Xia
2022 arXiv   pre-print
This study investigates weakly supervised image segmentation using loose bounding box supervision.  ...  It presents a multiple instance learning strategy based on polar transformation to assist image segmentation when loose bounding boxes are employed as supervision.  ...  [4] imposed a set of constraints on the network outputs based on the tightness prior of bounding boxes for image segmentation.  ... 
arXiv:2203.06000v1 fatcat:ctl2ksgzbjgd7hehxkkui4wkra

Accurate Cup-to-Disc Ratio Measurement with Tight Bounding Box Supervision in Fundus Photography [article]

Juan Wang, Bin Xia
2021 arXiv   pre-print
using only tight bounding box supervision.  ...  The weakly supervised image segmentation task is implemented based on generalized multiple instance learning formulation and smooth maximum approximation, and the bounding-box regression task outputs class-specific  ...  With bounding box tightness prior, a natural solution for CDR measurement in fundus images is to develop weakly supervised image segmentation (WSIS) method using tight bounding box supervision.  ... 
arXiv:2110.00943v1 fatcat:667flepf5fh77c75w4m6evzzim

Weakly- and Semi-supervised Panoptic Segmentation [chapter]

Qizhu Li, Anurag Arnab, Philip H. S. Torr
2018 Lecture Notes in Computer Science  
"Thing" classes are weakly-supervised with bounding boxes, and "stuff" with image-level tags.  ...  Furthermore, we present the first weakly-supervised results on Cityscapes for both semantic-and instance-segmentation.  ...  ., the EPSRC, Clarendon Fund, ERC grant ERC-2012-AdG 321162-HELIOS, EPRSRC grant Seebibyte EP/M013774/1 and EPSRC/MURI grant EP/N019474/1.  ... 
doi:10.1007/978-3-030-01267-0_7 fatcat:lqqibdtrynebhf3cjgjhj23voq

Amodal Segmentation through Out-of-Task and Out-of-Distribution Generalization with a Bayesian Model [article]

Yihong Sun, Adam Kortylewski, Alan Yuille
2021 arXiv   pre-print
The model is trained from non-occluded images using bounding box annotations and class labels only, but is applied to generalize out-of-task to object segmentation and to generalize out-of-distribution  ...  Our algorithm outperforms alternative methods that use the same supervision by a large margin, and even outperforms methods where annotated amodal segmentations are used during training, when the amount  ...  [19] uses box-level annotations to achieve instance segmentation by exploiting the bounding box tightness prior.  ... 
arXiv:2010.13175v3 fatcat:wvltbk7lbrewdi6zm4f4ehesku

Efficient Weakly-Supervised Object Detection with Pseudo Annotations

Qingsheng Yuan, Gang Sun, Jianming Liang, Biao Leng
2021 IEEE Access  
Although a fully-supervised object detector can be trained using annotations generated from the weakly-supervised object detector, the performance is still severely limited due to the low quality of mined  ...  The main barriers to the efficiency of WSOD are the ineffective data augmentations and inaccurate bounding box predictions.  ...  WEAKLY-SUPERVISED INSTANCE SEGMENTATION Compared with weakly-supervised object detection (WSOD) and weakly-supervised semantic segmentation (WSSS), a more exact kind of annotation, bounding boxes, has  ... 
doi:10.1109/access.2021.3099497 fatcat:w7kjwwdy6japlgop3ikwfl5vh4

Weakly- and Semi-Supervised Panoptic Segmentation [article]

Qizhu Li, Anurag Arnab, Philip H.S. Torr
2019 arXiv   pre-print
"Thing" classes are weakly-supervised with bounding boxes, and "stuff" with image-level tags.  ...  Furthermore, we present the first weakly-supervised results on Cityscapes for both semantic- and instance-segmentation.  ...  ., the EPSRC, Clarendon Fund, ERC grant ERC-2012-AdG 321162-HELIOS, EPRSRC grant Seebibyte EP/M013774/1 and EPSRC/MURI grant EP/N019474/1.  ... 
arXiv:1808.03575v3 fatcat:3hixm75un5ao7gkasdzi2o4ttq

TS$$^{2}$$2C: Tight Box Mining with Surrounding Segmentation Context for Weakly Supervised Object Detection [chapter]

Yunchao Wei, Zhiqiang Shen, Bowen Cheng, Honghui Shi, Jinjun Xiong, Jiashi Feng, Thomas Huang
2018 Lecture Notes in Computer Science  
This work provides a simple approach to discover tight object bounding boxes with only image-level supervision, called Tight box mining with Surrounding Segmentation Context (TS 2 C).  ...  TS 2 C leverages surrounding segmentation context derived from weakly-supervised segmentation to suppress such lowquality distracting candidates and boost the high-quality ones.  ...  In this case, we may need to develop effective instance-level semantic segmentation approaches in a weakly supervised manner.  ... 
doi:10.1007/978-3-030-01252-6_27 fatcat:hgglginri5fyde3nynxktkkmoa

TS2C: Tight Box Mining with Surrounding Segmentation Context for Weakly Supervised Object Detection [article]

Yunchao Wei and Zhiqiang Shen and Bowen Cheng and Honghui Shi and Jinjun Xiong and Jiashi Feng and Thomas Huang
2018 arXiv   pre-print
This work provides a simple approach to discover tight object bounding boxes with only image-level supervision, called Tight box mining with Surrounding Segmentation Context (TS2C).  ...  TS2C leverages surrounding segmentation context derived from weakly-supervised segmentation to suppress such low-quality distracting candidates and boost the high-quality ones.  ...  In this case, we may need to develop effective instance-level semantic segmentation approaches in a weakly supervised manner.  ... 
arXiv:1807.04897v1 fatcat:mosa5gnnbvcvpnfyqyq3rbe6eq

Cyclic Guidance for Weakly Supervised Joint Detection and Segmentation

Yunhang Shen, Rongrong Ji, Yan Wang, Yongjian Wu, Liujuan Cao
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
In this paper, we join weakly supervised object detection and segmentation tasks with a multi-task learning scheme for the first time, which uses their respective failure patterns to complement each other's  ...  In particular, we present an efficient and effective framework termed Weakly Supervised Joint Detection and Segmentation (WS-JDS).  ...  We use three mask extraction strategies: The first strategy uses the entire bounding boxes as the instance masks (BB). The second strategy fits a maximum ellipse on the bounding boxes (ELL).  ... 
doi:10.1109/cvpr.2019.00079 dblp:conf/cvpr/ShenJWWC19 fatcat:k7rkomxfwzchzpnw76iz7f4dsm

Weakly Supervised Instance Segmentation by Deep Community Learning [article]

Jaedong Hwang, Seohyun Kim, Jeany Son, Bohyung Han
2020 arXiv   pre-print
We present a weakly supervised instance segmentation algorithm based on deep community learning with multiple tasks.  ...  We address this problem by designing a unified deep neural network architecture, which has a positive feedback loop of object detection with bounding box regression, instance mask generation, instance  ...  Note that the instance segmentation accuracy of the detection-only model is given by using detected bounding boxes as segmentation masks.  ... 
arXiv:2001.11207v3 fatcat:b5wdslbtezf65ezansgzdkisai

Weakly Supervised Volumetric Segmentation via Self-taught Shape Denoising Model [article]

Qian He, Shuailin Li, Xuming He
2021 arXiv   pre-print
Weakly supervised segmentation is an important problem in medical image analysis due to the high cost of pixelwise annotation.  ...  To address this, we propose a novel weakly-supervised segmentation strategy capable of better capturing 3D shape prior in both model prediction and learning.  ...  For our loose bounding box, we generate each edge of the box with distance of 10-20 pixels to the corresponding tight bounding box edge.  ... 
arXiv:2104.13082v2 fatcat:gbkdduumrraldglhprpow7hp7q

Weakly Supervised Object Localization and Detection: A Survey [article]

Dingwen Zhang, Junwei Han, Gong Cheng, Ming-Hsuan Yang
2021 arXiv   pre-print
applications of the weakly supervised object localization and detection methods, and potential future directions to further promote the development of this research field.  ...  As an emerging and challenging problem in the computer vision community, weakly supervised object localization and detection plays an important role for developing new generation computer vision systems  ...  The predicted segmentation masks are used to mine object proposals with tight boxes, and fed into the online instance classifier refinement (OICR) network to learn weakly supervised object detector.  ... 
arXiv:2104.07918v1 fatcat:dwl6sjfzibdilnvjnrbifp4uke

BBAM: Bounding Box Attribution Map for Weakly Supervised Semantic and Instance Segmentation [article]

Jungbeom Lee, Jihun Yi, Chaehun Shin, Sungroh Yoon
2021 arXiv   pre-print
Weakly supervised segmentation methods using bounding box annotations focus on obtaining a pixel-level mask from each box containing an object.  ...  These areas constitute a bounding-box attribution map (BBAM), which identifies the target object in its bounding box and thus serves as pseudo ground-truth for weakly supervised semantic and instance segmentation  ...  [27] considered the effect of up to ±15% of label noise: we extend this to ±20%. The validity of the bounding box tightness priors used by Hsu et al.  ... 
arXiv:2103.08907v1 fatcat:7syqrg74yvei5bc3f35252tjku
« Previous Showing results 1 — 15 out of 1,430 results