Filters








19,515 Hits in 6.4 sec

A Convex Method for Locating Regions of Interest with Multi-instance Learning [chapter]

Yu-Feng Li, James T. Kwok, Ivor W. Tsang, Zhi-Hua Zhou
2009 Lecture Notes in Computer Science  
This can be accomplished with multi-instance learning techniques by treating each image as a bag of instances (regions).  ...  In content-based image retrieval (CBIR) and image screening, it is often desirable to locate the regions of interest (ROI) in the images automatically.  ...  highly competitive performance with the other state-of-the-art multi-instance learning methods. Fig. 1 shows some example images with the located ROIs.  ... 
doi:10.1007/978-3-642-04174-7_2 fatcat:qii4u26bmnaoxl5srbiqt4deza

C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection [article]

Fang Wan, Chang Liu, Wei Ke, Xiangyang Ji, Jianbin Jiao, Qixiang Ye
2019 arXiv   pre-print
In this paper, we introduce a continuation optimization method into MIL and thereby creating continuation multiple instance learning (C-MIL), with the intention of alleviating the non-convexity problem  ...  Weakly supervised object detection (WSOD) is a challenging task when provided with image category supervision but required to simultaneously learn object locations and object detectors.  ...  instance learning and thereby creating continuation multiple instance learning (C-MIL), with the purpose of alleviating the non-convexity problem in a systematic manner.  ... 
arXiv:1904.05647v1 fatcat:2kjkb6t65vdqnfchs7sye66yoi

C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection

Fang Wan, Chang Liu, Wei Ke, Xiangyang Ji, Jianbin Jiao, Qixiang Ye
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
In this paper, we introduce a continuation optimization method into MIL and thereby creating continuation multiple instance learning (C-MIL), with the intention of alleviating the non-convexity problem  ...  Weakly supervised object detection (WSOD) is a challenging task when provided with image category supervision but required to simultaneously learn object locations and object detectors.  ...  instance learning and thereby creating continuation multiple instance learning (C-MIL), with the purpose of alleviating the non-convexity problem in a systematic manner.  ... 
doi:10.1109/cvpr.2019.00230 dblp:conf/cvpr/WanLKJJY19 fatcat:iahos3poxfca3ktd5qzt567kfy

Weakly Supervised Cascaded Convolutional Networks [article]

Ali Diba, Vivek Sharma, Ali Pazandeh, Hamed Pirsiavash, Luc Van Gool
2016 arXiv   pre-print
The final stage of both architectures is a part of a convolutional neural network that performs multiple instance learning on proposals extracted in the previous stage(s).  ...  A new architecture of cascaded networks is proposed to learn a convolutional neural network (CNN) under such conditions.  ...  The authors would like to thank Nvidia for GPU donation.  ... 
arXiv:1611.08258v1 fatcat:cbdt7v5m2jdr7bypq4hvec23ru

Convex and Scalable Weakly Labeled SVMs [article]

Yu-Feng Li, Ivor W. Tsang, James T. Kwok, Zhi-Hua Zhou
2013 arXiv   pre-print
Experiments on three weakly labeled learning tasks, namely, (i) semi-supervised learning; (ii) multi-instance learning for locating regions of interest in content-based information retrieval; and (iii)  ...  This includes, for example, (i) semi-supervised learning where labels are partially known; (ii) multi-instance learning where labels are implicitly known; and (iii) clustering where labels are completely  ...  The MMC results are copied from their Multi-Instance Learning for Locating ROIs In this section, we evaluate the proposed method on multi-instance learning, with application to ROI-location in CBIR  ... 
arXiv:1303.1271v5 fatcat:3xqbttwx5rffjmifidls5tapfi

Non-negative matrix completion for action detection

Ehsan Adeli-Mosabbeb, Mahmood Fathy
2015 Image and Vision Computing  
This approach has a multi-label weakly supervised setting for activity detection, with a convex optimization procedure.  ...  In this paper, we present a method based on a non-negative matrix completion framework, that learns to label videos with activity classes, and localizes the activity of interest spatio-temporally throughout  ...  One of the works towards convex multiple instance learning was [30] , in which the authors propose a model based on matrix completion for MIL.  ... 
doi:10.1016/j.imavis.2015.04.006 fatcat:62ak7nkehvcwzpyta45xvd4sui

Multi-instance Methods for Partially Supervised Image Segmentation [chapter]

Andreas Müller, Sven Behnke
2012 Lecture Notes in Computer Science  
We formulate the problem of image segmentation as a multi-instance task on a given set of overlapping candidate segments.  ...  Using these candidate segments, we solve the multi-instance, multi-class problem using multi-instance kernels with an SVM.  ...  For each image, these segments are a set of overlapping, object-like regions, which serve as candidates for object locations.  ... 
doi:10.1007/978-3-642-28258-4_12 fatcat:a5knctennjbahcnqxbld7e33s4

Matrix Completion for Weakly-Supervised Multi-Label Image Classification

Ricardo Cabral, Fernando De la Torre, Joao Paulo Costeira, Alexandre Bernardino
2015 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Most methods for visual recognition are fully supervised, as they make use of bounding boxes or pixelwise segmentations to locate objects of interest.  ...  Compared to previous work, our proposed framework has three advantages: (1) Unlike existing solutions based on multiple-instance learning methods, our model is convex.  ...  MIL approaches consider images as bags with many instances denoting possible regions of interest.  ... 
doi:10.1109/tpami.2014.2343234 pmid:26353213 fatcat:yg2pxlmu7jgmtn4mqxgmpoq3rm

Multi-scale discriminative Region Discovery for Weakly-Supervised Object Localization [article]

Pei Lv, Haiyu Yu, Junxiao Xue, Junjin Cheng, Lisha Cui, Bing Zhou, Mingliang Xu, Yi Yang
2019 arXiv   pre-print
In this paper, we propose a simple yet effective multi-scale discriminative region discovery method to localize not only more integral objects but also as many as possible with only image-level class labels  ...  To mine more discriminative regions for the task of object localization, the multiple local maximum from the gradient weight maps are leveraged to generate the localization map with a parallel sliding  ...  [35] come up with a relaxed multiple-instance algorithm to translate multi-instance learning into a convex optimization problem. Oquab et al.  ... 
arXiv:1909.10698v1 fatcat:6vcy7a4qibdabjgg7dp4ep5mj4

Localized Image Retrieval Based on Interest Points

Meng Fanjie, Guo Baolong, Wu Xianxiang
2012 Procedia Engineering  
Then the normalized image is divided into a series of sector sub-regions with different area according to the distribution of interest points.  ...  This paper proposes a novel method for content-based image retrieval based on interest points. Interest points are detected from the scale and rotation normalized image.  ...  Acknowledgements This work is supported by the National Natural Science Foundation of China under Grants 61105066 and the Fundamental Research Funds for the Central Universities under Grants K50510040007  ... 
doi:10.1016/j.proeng.2012.01.496 fatcat:efloq3siirbk3hxhvytj3q3p3a

A probabilistic convex hull query tool

Zhou Zhao, Da Yan, Wilfred Ng
2012 Proceedings of the 15th International Conference on Extending Database Technology - EDBT '12  
Probabilistic convex hull is very useful for discovering the territory of imprecise data in such applications with a high confidence.  ...  We demonstrate two interesting results from studying the migration habit of one specific species and the correlation between species through probabilistic convex hull queries.  ...  The deterministic convex hull algorithms are not able to solve this problem, since it returns a convex hull with a lot of false positive regions because it considers all the possible sample locations.  ... 
doi:10.1145/2247596.2247668 dblp:conf/edbt/ZhaoYN12a fatcat:gei2si2ppfhsxipkuyjeagpv4y

Weakly Supervised Cascaded Convolutional Networks

Ali Diba, Vivek Sharma, Ali Pazandeh, Hamed Pirsiavash, Luc Van Gool
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
The final stage of both architectures is a part of a convolutional neural network that performs multiple instance learning on proposals extracted in the previous stage(s).  ...  A new architecture of cascaded networks is proposed to learn a convolutional neural network (CNN) under such conditions.  ...  The authors would like to thank Nvidia for GPU donation.  ... 
doi:10.1109/cvpr.2017.545 dblp:conf/cvpr/DibaSPPG17 fatcat:btei6ijyxbf4xc6x4hahnirs7u

Inferring Gene Interaction Networks from ISH Images via Kernelized Graphical Models [chapter]

Kriti Puniyani, Eric P. Xing
2012 Lecture Notes in Computer Science  
We propose a novel kernel-based graphical model learning algorithm, that is both convex and consistent.  ...  The algorithm uses multi-instance kernels to compute similarity between the expression patterns of different genes, and then minimizes the L1 regularized Bregman divergence to estimate a sparse gene interaction  ...  Multi-instance learning is a form of supervised learning, where instead of labeling each instance, a bag of instances is labeled.  ... 
doi:10.1007/978-3-642-33783-3_6 fatcat:rwcchcyv7zf4dn7hpa4rkiymoa

Small, Sparse, but Substantial: Techniques for Segmenting Small Agricultural Fields Using Sparse Ground Data [article]

Smit Marvaniya, Umamaheswari Devi, Jagabondhu Hazra, Shashank Mujumdar, Nitin Gupta
2020 arXiv   pre-print
Hence, in this paper, we present a multi-stage approach that uses a combination of machine learning and image processing techniques.  ...  In an evaluation using high-resolution imagery, we show that our approach has a high F-Score of 0.84 in areas with large fields and reasonable accuracy with an F-Score of 0.73 in areas with small fields  ...  Identifying fields from images is a special instance of the image segmentation problem, for which several traditional non-learning based algorithms Zhu et al. (2015) as well as deep-learning models  ... 
arXiv:2005.01947v1 fatcat:dcwnn72uzrerfmhmp7mqsrmfzy

M4L: Maximum margin Multi-instance Multi-cluster Learning for scene modeling

Tianzhu Zhang, Si Liu, Changsheng Xu, Hanqing Lu
2013 Pattern Recognition  
Therefore, blocks can be grouped as a Multi-instance Multi-cluster Learning (MIMCL) problem, and a novel Maximum Margin Multi-instance Multi-cluster Learning (M 4 L) algorithm is proposed.  ...  To avoid processing a difficult optimization problem, M 4 L is further relaxed and solved by making use of a combination of the Cutting Plane method and Constrained Concave-Convex Procedure (CCCP).  ...  For video scene understanding, we propose a maximum margin multi-instance multi-cluster learning (M 4 L) method, which is closely related to the learning frameworks of multi-instance learning [30] , multi-label  ... 
doi:10.1016/j.patcog.2013.02.018 fatcat:s6yvu26mtbapxnkegeuqkigpme
« Previous Showing results 1 — 15 out of 19,515 results