{"id":358,"date":"2015-07-30T17:47:40","date_gmt":"2015-07-30T17:47:40","guid":{"rendered":"http:\/\/sicv.activearchives.org\/logbook\/?p=358"},"modified":"2016-01-07T10:25:08","modified_gmt":"2016-01-07T10:25:08","slug":"one-millisecond-face-alignment-with-an-ensemble-of-regression-trees","status":"publish","type":"post","link":"https:\/\/sicv.activearchives.org\/logbook\/one-millisecond-face-alignment-with-an-ensemble-of-regression-trees\/","title":{"rendered":"&#8220;One Millisecond Face Alignment with an Ensemble of Regression Trees&#8221;"},"content":{"rendered":"<blockquote><p>This paper addresses the problem of Face Alignment for<br \/>\na single image. We show how an ensemble of regression<br \/>\ntrees can be used to estimate the face\u2019s landmark positions<br \/>\ndirectly from a sparse subset of pixel intensities, achieving<br \/>\nsuper-realtime performance with high quality predictions.<br \/>\nWe present a general framework based on gradient boosting<br \/>\nfor learning an ensemble of regression trees that optimizes<br \/>\nthe sum of square error loss and naturally handles missing<br \/>\nor partially labelled data. We show how using appropriate<br \/>\npriors exploiting the structure of image data helps with ef-<br \/>\nficient feature selection. Different regularization strategies<br \/>\nand its importance to combat overfitting are also investi-<br \/>\ngated. In addition, we analyse the effect of the quantity of<br \/>\ntraining data on the accuracy of the predictions and explore<br \/>\nthe effect of data augmentation using synthesized data.<\/p><\/blockquote>\n<p>From <a href=\"http:\/\/www.csc.kth.se\/~vahidk\/papers\/KazemiCVPR14.pdf\">One Millisecond Face Alignment with an Ensemble of Regression Trees<\/a> Vahid Kazemi and Josephine Sullivan; KTH, Royal Institute of Technology<br \/>\nComputer Vision and Active Perception Lab<br \/>\nTeknikringen 14, Stockholm, Sweden<\/p>\n<p><a href=\"http:\/\/matthewearl.github.io\/2015\/07\/28\/switching-eds-with-python\/\"><img loading=\"lazy\" class=\"alignnone size-full wp-image-362\" src=\"http:\/\/sicv.activearchives.org\/logbook\/wp-content\/uploads\/2015\/07\/Matthewearl_faceswap.png\" alt=\"Matthewearl_faceswap\" width=\"267\" height=\"256\" \/><\/a><\/p>\n<p>Cited by <a href=\"http:\/\/matthewearl.github.io\/2015\/07\/28\/switching-eds-with-python\/\">this blog post<\/a> which uses the technique, among other things to match and then project one face onto another.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>This paper addresses the problem of Face Alignment for a single image. We show how an ensemble of regression trees can be used to estimate the face\u2019s landmark positions directly from a sparse subset of pixel intensities, achieving super-realtime performance with high quality predictions. We present a general framework based on gradient boosting for learning [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":360,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[5],"tags":[51],"_links":{"self":[{"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/posts\/358"}],"collection":[{"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/comments?post=358"}],"version-history":[{"count":5,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/posts\/358\/revisions"}],"predecessor-version":[{"id":445,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/posts\/358\/revisions\/445"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/media\/360"}],"wp:attachment":[{"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/media?parent=358"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/categories?post=358"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/sicv.activearchives.org\/logbook\/wp-json\/wp\/v2\/tags?post=358"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}