For object classification, soft assignment (SA) is capable
\nof improving the bag-of-visual-words (BoVW) model and
\nhas the advantages in conceptual simplicity. However, the
\nperformance of soft assignment is inferior to those recently
\ndeveloped encoding schemes. In this paper, we propose a
\nnovel scheme called barycentric coordinates based soft assignment
\n(BCSA) for the classification of object images.
\nWhile maintaining conceptual simplicity, this scheme will be
\nshown to outperform most of the existing encoding schemes,
\nincluding sparse and local coding schemes. Furthermore, with
\nonly single-scale features, it is able to achieve comparable or
\neven better performance to current state-of-the-art Fisher kernel
\n(FK) encoding scheme. In particular, the proposed BCSA
\nscheme enjoys the following properties: 1) preservation of
\nlinear order precision for encoding which makes BCSA robust
\nto linear transform distortions; 2) inheriting naturally
\nthe visual word uncertainty which leads to a more expressive
\nmodel; 3) generating linear classifiable codes that can be
\nlearned with significant less computational cost and storage.
\nExtensive experiments based on widely used Caltech-101 and
\nCaltech-256 datasets have been carried out to show its effectiveness
\nof the proposed BCSA scheme in both performance
\nand simplicity.<\/p>\n","protected":false},"excerpt":{"rendered":"
For object classification, soft assignment (SA) is capable of improving the bag-of-visual-words (BoVW) model and has the advantages in conceptual simplicity. However, the performance of soft assignment is inferior to those recently developed encoding schemes. In this paper, we propose a novel scheme called barycentric coordinates based soft assignment (BCSA) for the classification of object […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":null,"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"IEEE International Conference on Multimedia & Expo Workshops","msr_doi":"","msr_arxiv_id":"","msr_s2_paper_id":"","msr_mag_id":"","msr_pubmed_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_original_fields_of_study":"","msr_release_tracker_id":"","msr_s2_match_type":"","msr_citation_count_updated":"","msr_published_date":"2016-07-29","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_s2_pdf_url":"","msr_year":0,"msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":0,"msr_s2_match_confidence":0,"msr_microsoftintellectualproperty":true,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13562],"msr-publication-type":[193716],"msr-publisher":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-267888","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-computer-vision","msr-locale-en_us"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2016-07-29","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"267894","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"file","title":"00-2016-93-icme2016-workshop","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2016\/07\/00-2016-93-icme2016-workshop.pdf","id":267894,"label_id":0}],"msr_related_uploader":"","msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"","msr_influential_citations":0,"msr_reference_count":0,"msr_arxiv_id":"","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Tao Wei","user_id":0,"rest_url":false},{"type":"text","value":"Chang Wen Chen","user_id":0,"rest_url":false},{"type":"user_nicename","value":"chw","user_id":31440,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=chw"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[285614,212093],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"inproceedings","related_content":{"projects":[{"ID":285614,"post_title":"Image2Text","post_name":"image2text-2","post_type":"msr-project","post_date":"2016-08-30 20:14:19","post_modified":"2017-06-15 14:52:31","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/image2text-2\/","post_excerpt":"We study the problem of image captioning, i.e., automatically describing an image by a sentence. This is a challenging problem, since different from other computer vision tasks such as image classi\ufb01cation and object detection, image captioning requires not only understanding the image, but also the knowledge of natural language. We formulate this problem as a multimodal translation task, and develop novel algorithms to solve this problem.","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/285614"}]}},{"ID":212093,"post_title":"Image\/Video Understanding and Analysis","post_name":"image2text","post_type":"msr-project","post_date":"2016-01-25 01:52:15","post_modified":"2017-06-15 14:48:39","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/image2text\/","post_excerpt":"We target at the core problems in image\/video understanding and analysis, such as image recognition, image segmentation, image captioning, image parsing, object detection, and video segmentation.","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/212093"}]}}]},"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/267888","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/267888\/revisions"}],"predecessor-version":[{"id":531635,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/267888\/revisions\/531635"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=267888"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=267888"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=267888"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=267888"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=267888"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=267888"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=267888"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=267888"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=267888"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=267888"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=267888"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=267888"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=267888"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}