Joint Dictionary and Classifier learning for Categorization of Images using a Max-margin Framework. Lobel, H., Vidal, R., Mery, D., & Soto., A. In 6th Pacific-Rim Symposium on Image and Video Technology, PSIVT, 2013. Paper abstract bibtex 7 downloads The Bag-of-Visual-Words (BoVW) model is a popular approach for visual recognition. Used successfully in many different tasks, simplicity and good performance are the main reasons for its popularity. The central aspect of this model, the visual dictionary, is used to build mid-level representations based on low level image descriptors. Classifiers are then trained using these mid-level representations to perform categorization. While most works based on BoVW models have been focused on learning a suitable dictionary or on proposing a suitable pooling strategy, little effort has been devoted to explore and improve the coupling between the dictionary and the top-level classifiers, in order to gen- erate more discriminative models. This problem can be highly complex due to the large dictionary size usually needed by these methods. Also, most BoVW based systems usually perform multiclass categorization using a one-vs-all strat- egy, ignoring relevant correlations among classes. To tackle the previous issues, we propose a novel approach that jointly learns dictionary words and a proper top- level multiclass classifier. We use a max-margin learning framework to minimize a regularized energy formulation, allowing us to propagate labeled information to guide the commonly unsupervised dictionary learning process. As a result we produce a dictionary that is more compact and discriminative. We test our method on several popular datasets, where we demonstrate that our joint optimization strategy induces a word sharing behavior among the target classes, being able to achieve state-of-the-art performance using far less visual words than previous approaches.
@InProceedings{ lobel-b:etal:2013,
author = {H. Lobel and R. Vidal and D. Mery and A. Soto.},
title = {Joint Dictionary and Classifier learning for
Categorization of Images using a Max-margin Framework},
booktitle = {6th Pacific-Rim Symposium on Image and Video Technology,
PSIVT},
year = {2013},
abstract = {The Bag-of-Visual-Words (BoVW) model is a popular approach
for visual recognition. Used successfully in many different
tasks, simplicity and good performance are the main reasons
for its popularity. The central aspect of this model, the
visual dictionary, is used to build mid-level
representations based on low level image descriptors.
Classifiers are then trained using these mid-level
representations to perform categorization. While most works
based on BoVW models have been focused on learning a
suitable dictionary or on proposing a suitable pooling
strategy, little effort has been devoted to explore and
improve the coupling between the dictionary and the
top-level classifiers, in order to gen- erate more
discriminative models. This problem can be highly complex
due to the large dictionary size usually needed by these
methods. Also, most BoVW based systems usually perform
multiclass categorization using a one-vs-all strat- egy,
ignoring relevant correlations among classes. To tackle the
previous issues, we propose a novel approach that jointly
learns dictionary words and a proper top- level multiclass
classifier. We use a max-margin learning framework to
minimize a regularized energy formulation, allowing us to
propagate labeled information to guide the commonly
unsupervised dictionary learning process. As a result we
produce a dictionary that is more compact and
discriminative. We test our method on several popular
datasets, where we demonstrate that our joint optimization
strategy induces a word sharing behavior among the target
classes, being able to achieve state-of-the-art performance
using far less visual words than previous approaches. },
url = {http://saturno.ing.puc.cl/media/papers_alvaro/Hans-PSIVT-13.pdf}
}
Downloads: 7
{"_id":{"_str":"53427a470e946d920a0018df"},"__v":1,"authorIDs":["24TkaxcPXqc7t8oo2","32ZR23o2BFySHbtQK","3ear6KFZSRqbj6YeT","4Pq6KLaQ8jKGXHZWH","5456ff2b8b01c819300000f6","54578d9a2abc8e9f370004f0","5de77464021482de01000070","5e126ca5a4cabfdf01000053","5e158f76f1f31adf01000118","5e16174bf67f7dde010003ad","5e1c779a92587bde0100013e","5e1f631ae8f5ddde010000eb","5e1f7182e8f5ddde010001ff","5e221d6c71dcf8df0100002b","5e253d35561b8fde0100008c","5e26da3642065ede01000066","5e28810c67e11edf0100008d","5e3acefaf2a00cdf010001c8","5e533ea853de8dde01000042","5e54fee296ed20df0100003d","5e62c3aecb259cde010000f9","5e65830c6e5f4cf3010000e7","5e666dfc46e828de010002c9","6cMBYieMJhf6Nd58M","6w6sGsxYSK2Quk6yZ","7xDcntrrtC62vkWM5","ARw5ReidxxZii9TTZ","BjzM7QpRCG7uCF7Zf","DQ4JRTTWkvKXtCNCp","DR8QRotRb5C33T8Rb","FDHBz8WWdRXnWmR7h","GbYBJvxugXMriQwbi","HhRoRmBvwWfD4oLyK","JFk6x26H6LZMoht2n","JvArGGu5qM6EvSCvB","LpqQBhFH3PxepH9KY","MT4TkSGzAp69M3dGt","NKHZCaBn3zKAa5ocL","Pxb9bPepYv65oXxi9","QFECgvB5v2i4j2Qzs","RKv56Kes3h6FwEa55","Rb9TkQ3KkhGAaNyXq","RdND8NxcJDsyZdkcK","SpKJ5YujbHKZnHc4v","TSRdcx4bbYKqcGbDg","W8ogS2GJa6sQKy26c","WTi3X2fT8dzBN5d8b","WfZbctNQYDBaiYW6n","XZny8xuqwfoxzhBCB","Xk2Q5qedS5MFHvjEW","ayym4ZHJzF7jQcz8Q","bbARiTJLYS79ZMFbk","cBxsyeZ37EucQeBYK","cFyFQps7W3Sa2Wope","d6Qqa2JYQxcTSLaGE","dGRBfr8zhMmbwK6eP","dmtweywdYxkx3KjDt","eRLgwkrEk7T7Lmzmf","fMYSCX8RMZap548vv","fN7f2faCKTWJvL9rb","g6iKCQCFnJgKYYHaP","h2WsvS5g4tT8oqpZH","h2hTcQYuf2PB3oF8t","h83jBvZYJPJGutQrs","jAtuJBcGhng4Lq2Nd","pMoo2gotJcdDPwfrw","q5Zunk5Y2ruhw5vyq","rzNGhqxkbt2MvGY29","syJh4acsFmZGsFfQS","uC8ATA8AfngWpYLBq","uoJ7BKv28Q6TtPmPp","vMiJzqEKCsBxBEa3v","vQE6iTPpjxpuLip2Z","wQDRsDjhgpMJDGxWX","wbNg79jvDpzX9zHLK","wk86BgRiooBjy323E","zCbPxKnQGgDHiHMWn","zf9HENjsAzdWLMDAu"],"author_short":["Lobel, H.","Vidal, R.","Mery, D.","Soto., A."],"bibbaseid":"lobel-vidal-mery-soto-jointdictionaryandclassifierlearningforcategorizationofimagesusingamaxmarginframework-2013","bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"firstnames":["H."],"propositions":[],"lastnames":["Lobel"],"suffixes":[]},{"firstnames":["R."],"propositions":[],"lastnames":["Vidal"],"suffixes":[]},{"firstnames":["D."],"propositions":[],"lastnames":["Mery"],"suffixes":[]},{"firstnames":["A."],"propositions":[],"lastnames":["Soto."],"suffixes":[]}],"title":"Joint Dictionary and Classifier learning for Categorization of Images using a Max-margin Framework","booktitle":"6th Pacific-Rim Symposium on Image and Video Technology, PSIVT","year":"2013","abstract":"The Bag-of-Visual-Words (BoVW) model is a popular approach for visual recognition. Used successfully in many different tasks, simplicity and good performance are the main reasons for its popularity. The central aspect of this model, the visual dictionary, is used to build mid-level representations based on low level image descriptors. Classifiers are then trained using these mid-level representations to perform categorization. While most works based on BoVW models have been focused on learning a suitable dictionary or on proposing a suitable pooling strategy, little effort has been devoted to explore and improve the coupling between the dictionary and the top-level classifiers, in order to gen- erate more discriminative models. This problem can be highly complex due to the large dictionary size usually needed by these methods. Also, most BoVW based systems usually perform multiclass categorization using a one-vs-all strat- egy, ignoring relevant correlations among classes. To tackle the previous issues, we propose a novel approach that jointly learns dictionary words and a proper top- level multiclass classifier. We use a max-margin learning framework to minimize a regularized energy formulation, allowing us to propagate labeled information to guide the commonly unsupervised dictionary learning process. As a result we produce a dictionary that is more compact and discriminative. We test our method on several popular datasets, where we demonstrate that our joint optimization strategy induces a word sharing behavior among the target classes, being able to achieve state-of-the-art performance using far less visual words than previous approaches. ","url":"http://saturno.ing.puc.cl/media/papers_alvaro/Hans-PSIVT-13.pdf","bibtex":"@InProceedings{\t lobel-b:etal:2013,\n author\t= {H. Lobel and R. Vidal and D. Mery and A. Soto.},\n title\t\t= {Joint Dictionary and Classifier learning for\n\t\t Categorization of Images using a Max-margin Framework},\n booktitle\t= {6th Pacific-Rim Symposium on Image and Video Technology,\n\t\t PSIVT},\n year\t\t= {2013},\n abstract\t= {The Bag-of-Visual-Words (BoVW) model is a popular approach\n\t\t for visual recognition. Used successfully in many different\n\t\t tasks, simplicity and good performance are the main reasons\n\t\t for its popularity. The central aspect of this model, the\n\t\t visual dictionary, is used to build mid-level\n\t\t representations based on low level image descriptors.\n\t\t Classifiers are then trained using these mid-level\n\t\t representations to perform categorization. While most works\n\t\t based on BoVW models have been focused on learning a\n\t\t suitable dictionary or on proposing a suitable pooling\n\t\t strategy, little effort has been devoted to explore and\n\t\t improve the coupling between the dictionary and the\n\t\t top-level classifiers, in order to gen- erate more\n\t\t discriminative models. This problem can be highly complex\n\t\t due to the large dictionary size usually needed by these\n\t\t methods. Also, most BoVW based systems usually perform\n\t\t multiclass categorization using a one-vs-all strat- egy,\n\t\t ignoring relevant correlations among classes. To tackle the\n\t\t previous issues, we propose a novel approach that jointly\n\t\t learns dictionary words and a proper top- level multiclass\n\t\t classifier. We use a max-margin learning framework to\n\t\t minimize a regularized energy formulation, allowing us to\n\t\t propagate labeled information to guide the commonly\n\t\t unsupervised dictionary learning process. As a result we\n\t\t produce a dictionary that is more compact and\n\t\t discriminative. We test our method on several popular\n\t\t datasets, where we demonstrate that our joint optimization\n\t\t strategy induces a word sharing behavior among the target\n\t\t classes, being able to achieve state-of-the-art performance\n\t\t using far less visual words than previous approaches. },\n url\t\t= {http://saturno.ing.puc.cl/media/papers_alvaro/Hans-PSIVT-13.pdf}\n}\n\n","author_short":["Lobel, H.","Vidal, R.","Mery, D.","Soto., A."],"key":"lobel-b:etal:2013","id":"lobel-b:etal:2013","bibbaseid":"lobel-vidal-mery-soto-jointdictionaryandclassifierlearningforcategorizationofimagesusingamaxmarginframework-2013","role":"author","urls":{"Paper":"http://saturno.ing.puc.cl/media/papers_alvaro/Hans-PSIVT-13.pdf"},"metadata":{"authorlinks":{"mery, d":"https://domingomery.ing.puc.cl/testing/","soto, a":"https://asoto.ing.puc.cl/publications/"}},"downloads":7},"bibtype":"inproceedings","biburl":"https://raw.githubusercontent.com/ialab-puc/ialab.ing.puc.cl/master/pubs.bib","downloads":7,"keywords":[],"search_terms":["joint","dictionary","classifier","learning","categorization","images","using","max","margin","framework","lobel","vidal","mery","soto."],"title":"Joint Dictionary and Classifier learning for Categorization of Images using a Max-margin Framework","year":2013,"dataSources":["3YPRCmmijLqF4qHXd","sg6yZ29Z2xB5xP79R","sj4fjnZAPkEeYdZqL","m8qFBfFbjk9qWjcmJ","QjT2DEZoWmQYxjHXS","68BuKygEnwqbDeD59"]}