We present a convolutional neural network (CNN) based solution for
\nmodeling physically plausible spatially varying surface
\nreflectance functions (SVBRDF) from a single photograph of a
\nplanar material sample under unknown natural
\nillumination. Gathering a sufficiently large set of labeled
\ntraining pairs consisting of photographs of SVBRDF samples and
\ncorresponding reflectance parameters, is a difficult and arduous
\nprocess. To reduce the amount of required labeled training data,
\nwe propose to leverage the appearance information embedded in
\nunlabeled images of spatially varying materials to self-augment
\nthe training process. Starting from a coarse network obtained from
\na small set of labeled training pairs, we estimate provisional
\nmodel parameters for each unlabeled training exemplar. Given this
\nprovisional reflectance estimate, we then synthesize a novel
\ntemporary labeled training pair by rendering the exact
\ncorresponding image under a new lighting condition. After refining
\nthe network using these additional training samples, we
\nre-estimate the provisional model parameters for the unlabeled
\ndata and repeat the self-augmentation process until convergence.
\nWe demonstrate the efficacy of the proposed network structure on
\nspatially varying wood, metal, and plastics, as well as thoroughly
\nvalidate the effectiveness of the self-augmentation training
\nprocess.<\/p>\n","protected":false},"excerpt":{"rendered":"
We present a convolutional neural network (CNN) based solution for modeling physically plausible spatially varying surface reflectance functions (SVBRDF) from a single photograph of a planar material sample under unknown natural illumination. Gathering a sufficiently large set of labeled training pairs consisting of photographs of SVBRDF samples and corresponding reflectance parameters, is a difficult and […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"footnotes":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13551],"msr-publication-type":[193715],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-417845","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-graphics-and-multimedia","msr-locale-en_us"],"msr_publishername":"ACM","msr_edition":"ACM Transactions on Graphics","msr_affiliation":"","msr_published_date":"2017-07-30","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"ACM Transactions on Graphics","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":"http:\/\/YToxOntpOjA7YToxOntzOjM6InVybCI7czozNToiaHR0cDovL21zcmFpZy5pbmZvL35zYW5ldC9zYW5ldC5odG0iO319","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"417851","msr_publicationurl":"http:\/\/msraig.info\/~sanet\/sanet.htm","msr_doi":"","msr_publication_uploader":[{"type":"file","title":"sanet","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2017\/07\/sanet.pdf","id":417851,"label_id":0},{"type":"url","title":"http:\/\/msraig.info\/~sanet\/sanet.htm","viewUrl":false,"id":false,"label_id":0}],"msr_related_uploader":"","msr_attachments":[{"id":0,"url":"http:\/\/msraig.info\/~sanet\/sanet.htm"}],"msr-author-ordering":[{"type":"text","value":"Xiao Li","user_id":0,"rest_url":false},{"type":"user_nicename","value":"yuedong","user_id":35060,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=yuedong"},{"type":"text","value":"Pieter Peers","user_id":0,"rest_url":false},{"type":"user_nicename","value":"xtong","user_id":34929,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=xtong"}],"msr_impact_theme":[],"msr_research_lab":[199560],"msr_event":[],"msr_group":[144710],"msr_project":[],"publication":[],"video":[],"download":[],"msr_publication_type":"article","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/417845"}],"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":1,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/417845\/revisions"}],"predecessor-version":[{"id":417848,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/417845\/revisions\/417848"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=417845"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=417845"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=417845"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=417845"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=417845"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=417845"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=417845"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=417845"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=417845"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=417845"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=417845"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=417845"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=417845"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=417845"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=417845"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}