Filters








54,667 Hits in 3.5 sec

Unsupervised image ranking

Eva Hörster, Malcolm Slaney, Marc'Aurelio Ranzato, Kilian Weinberger
2009 Proceedings of the First ACM workshop on Large-scale multimedia retrieval and mining - LS-MMRM '09  
In the paper, we propose and test an unsupervised approach for image ranking. Prior solutions are based on image content and the similarity graph connecting images.  ...  Our approach is unsupervised and allows for various feature modalities. We demonstrate the effectiveness of our approach using both visual-content-based and tag-based features.  ...  Figure 5 : 5 Mean scores for the top-15 images over all categories for our baseline method (random), a text-based search engine, and 4 different variants of the unsupervised ranking algorithm.  ... 
doi:10.1145/1631058.1631074 dblp:conf/mm/HorsterSRW09 fatcat:anvxwshxnbfvhbochtjezngjbm

Unsupervised Rank Fusion for Diverse Image Metasearch

José Solenir L. Figuerêdo, Rodrigo Tripodi Calumby
2019 Anais do Simpósio Brasileiro de Banco de Dados (SBBD)  
For a given query and a set of images ranked lists retrieved from multiple search engines, the metasearch technique aims at combining these lists to build an unified ranking with improved relevance.  ...  Rank aggregation is an approach that has been widely used to support this task. This paper investigates the use of rank aggregation methods in the metasearch scenario for diverse image retrieval.  ...  This work experimentally investigates the effectiveness of several rank aggregation methods for metasearch in the context of diverse image retrieval.  ... 
doi:10.5753/sbbd.2019.8834 dblp:conf/sbbd/FigueredoC19 fatcat:z5u7vicnbbforkna4y5xqgs7xu

Unsupervised Rank-Preserving Hashing for Large-Scale Image Retrieval [article]

Svebor Karaman, Xudong Lin, Xuefeng Hu, Shih-Fu Chang
2019 arXiv   pre-print
We propose an unsupervised hashing method which aims to produce binary codes that preserve the ranking induced by a real-valued representation.  ...  Such compact hash codes enable the complete elimination of real-valued feature storage and allow for significant reduction of the computation complexity and storage cost of large-scale image retrieval  ...  Unsupervised methods, like ours, usually assume the rank in the original space is the rank we want to preserve.  ... 
arXiv:1903.01545v1 fatcat:6nbdlwzf5vds5nz6uwt5jisqiq

Unsupervised Ensemble Ranking: Application to Large-Scale Image Retrieval

Jung-Eun Lee, Rong Jin, Anil K. Jain
2010 2010 20th International Conference on Pattern Recognition  
An unsupervised algorithm is developed to learn the weights for fusing the rankings from multiple bag-of-words models.  ...  We circumvent this performance loss by an ensemble ranking approach in which rankings from multiple bag-of-words models are combined to obtain more accurate retrieval results.  ...  This unsupervised ensemble ranking problem is formulated below.  ... 
doi:10.1109/icpr.2010.950 dblp:conf/icpr/LeeJJ10 fatcat:ufrh77hswnbadfjxuoxty4jv5m

A hybrid unsupervised image re-ranking approach with latent topic contents

Lei Zhang, Piji Li, Jun Ma
2010 Proceedings of the ACM International Conference on Image and Video Retrieval - CIVR '10  
In this paper, our objective is to re-rank initial search results from these image search engines to improve user experience.  ...  We present an evaluation of pLSA and LDA as dimension reduction approach for the task of web image re-ranking and discuss the benefits of integrating topic distribution features.  ...  Figure 2 : 2 Top 21 results of a Google image search query for 'flag'. Figure 3 : 3 The flowchart of our hybrid unsupervised image re-ranking approach.  ... 
doi:10.1145/1816041.1816088 dblp:conf/civr/ZhangLM10 fatcat:3ztjweurvrd5xcfneoadhu23bu

Combination of Clustering and Ranking Techniques for Unsupervised Band Selection of Hyperspectral Images

Aloke Datta, Susmita Ghosh, Ashish Ghosh
2015 IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing  
Index Terms-Band clustering, band ranking, capacitory discrimination (CD), hyperspectral imagery, unsupervised band selection.  ...  An unsupervised band selection method is proposed in this article. It is a three-step procedure. In the first step, characteristics (attributes) of the bands are found out.  ...  Combination of Clustering and Ranking Techniques for Unsupervised Band Selection of Hyperspectral Images I.  ... 
doi:10.1109/jstars.2015.2428276 fatcat:cri576zberanbonito6o5zvaza

Unsupervised Selective Rank Fusion for Content-based Image Retrieval

Lucas Pascotti Valem, Daniel Carlos Guimarães Pedronette
2020 Anais do Concurso de Teses e Dissertações da SBC (CTD-SBC 2020)   unpublished
This work proposes three novel methods for selecting and combining ranked lists by estimating their effectiveness in an unsupervised way.  ...  The CBIR (Content-Based Image Retrieval) systems are one of the main solutions for image retrieval tasks.  ...  In this Master's work, three unsupervised methods are proposed for selection and fusion of ranked lists.  ... 
doi:10.5753/ctd.2020.11370 fatcat:mvfhzobst5edhimniasswtibsi

Unsupervised Vehicle Re-Identification via Self-supervised Metric Learning using Feature Dictionary [article]

Jongmin Yu, Hyeontaek Oh
2021 arXiv   pre-print
Pair-wise similarity, relative-rank consistency, and adjacent feature distribution similarity are jointly considered to find images that may belong to the same vehicle of a given probe image.  ...  The key challenge of unsupervised vehicle re-identification (Re-ID) is learning discriminative features from unlabelled vehicle images.  ...  methods, including different unsupervised Re-ID domains such as person Re-ID, in terms of rank-1 and rank-5 accuracies and mAP on the VeRi-776 dataset  ... 
arXiv:2103.02250v1 fatcat:pzoxdn5jirhmhmbdmkjelwejxa

Exploiting Robust Unsupervised Video Person Re-identification [article]

Xianghao Zang, Ge Li, Wei Gao, Xiujun Shu
2021 arXiv   pre-print
Features from these two modules are fused to form a robust feature representation for each input image.  ...  To improve the performance stability for unsupervised video reID, this paper introduces a general scheme fusing part models and unsupervised learning.  ...  The global Cyclic Ranking Consistency (CRC) [30] is employed to explore the relationship between images from different cameras.  ... 
arXiv:2111.05170v2 fatcat:u4s6xx4cubdorgn2n4wblprtje

Unsupervised Region Attention Network for Person Re-identification

Chenrui Zhang, Yangxu Wu, Tao Lei
2019 IEEE Access  
Existing unsupervised person Re-Id models learn global features of pedestrian from whole images or several constant patches.  ...  weights on images.  ...  Then, we sort the gallery images as a ranking list by their similarities in descending order.  ... 
doi:10.1109/access.2019.2953280 fatcat:3krqamxb2nddfo22fijpte5aom

A semi-supervised learning algorithm for relevance feedback and collaborative image retrieval

Daniel Carlos Guimarães Pedronette, Rodrigo T. Calumby, Ricardo da S. Torres
2015 EURASIP Journal on Image and Video Processing  
Later, an unsupervised learning step is performed with the objective of extracting useful information from the intrinsic dataset structure.  ...  Similar results are also observed in experiments considering multimodal image retrieval tasks.  ...  Unsupervised pairwise recommendation The Pairwise Recommendation [9] algorithm consists in an unsupervised image re-ranking method proposed for image retrieval tasks.  ... 
doi:10.1186/s13640-015-0081-6 fatcat:uinx4wos7fab3o2gnwywdlrq7a

Unsupervised Attention Based Instance Discriminative Learning for Person Re-Identification [article]

Kshitij Nikhal, Benjamin S. Riggan
2020 arXiv   pre-print
Therefore, we propose an unsupervised framework for person re-identification which is trained in an end-to-end manner without any pre-training.  ...  However, since the data requirements---including the degree of data curations---are becoming increasingly complex and laborious, there is a critical need for unsupervised methods that are robust to large  ...  Using the Market1501 dataset, Table 1 shows the rank-1, rank-5, and rank-10 Re-ID accuracy and mAP under the unsupervised scenario with pre-trained weights.  ... 
arXiv:2011.01888v1 fatcat:nhkjhbrejnb5rhs5d4rknvwa6u

Fully Unsupervised Person Re-identification viaSelective Contrastive Learning [article]

Bo Pang, Deming Zhai, Junjun Jiang, Xianming Liu
2021 arXiv   pre-print
Person re-identification (ReID) aims at searching the same identity person among images captured by various cameras.  ...  Representation learning plays a critical role in unsupervised person ReID. In this work, we propose a novel selective contrastive learning framework for unsupervised feature learning.  ...  Compared with fully unsupervised method MMCL, we achieve 6.0% and 2.1% improvement on rank-1 and mAP respectively.  ... 
arXiv:2010.07608v2 fatcat:c4zqhhd2wvex7ijmmvqh667a4m

Unsupervised Visual Object Categorisation with BoF and Spatial Matching [chapter]

Teemu Kinnunen, Jukka Lankinen, Joni-Kristian Kämäräinen, Lasse Lensu, Heikki Kälviäinen
2013 Lecture Notes in Computer Science  
In this work, we propose a novel unsupervised image categorisation method which uses the BoF to find initial matches for each image (pre-filter) and then refines and ranks them using spatial matching of  ...  The ultimate challenge of image categorisation is unsupervised object discovery, where the selection of categories and the assignments of given images to these categories are performed automatically.  ...  Our main contribution in this work is a novel method for the unsupervised image categorisation using the BoF as a pre-processing stage to rank image matches, and then a random sampling based spatial matching  ... 
doi:10.1007/978-3-642-38886-6_37 fatcat:3crnpnwcvradxlgydwfwb23l2m

Unsupervised Person Re-identification via Softened Similarity Learning [article]

Yutian Lin, Lingxi Xie, Yu Wu, Chenggang Yan, Qi Tian
2020 arXiv   pre-print
Experiments on two image-based and video-based datasets demonstrate state-of-the-art performance under the unsupervised re-ID setting.  ...  This paper studies the unsupervised setting of re-ID, which does not require any labeled information and thus is freely deployed to new scenarios.  ...  Methods Setting Market-1501 DukeMTMC-reID rank-1 rank-5 rank-10 mAP rank-1 rank-5 rank-10 mAP OIM [33] Unsupervised Table 1 .  ... 
arXiv:2004.03547v1 fatcat:7cs6jjfz2jgzxljkisehidm4w4
« Previous Showing results 1 — 15 out of 54,667 results