000199706 001__ 199706
000199706 005__ 20190316235925.0
000199706 037__ $$aREP_WORK
000199706 245__ $$aOn Rendering Synthetic Images for Training an Object Detector
000199706 269__ $$a2014
000199706 260__ $$c2014
000199706 300__ $$a20
000199706 336__ $$aReports
000199706 520__ $$aWe propose a novel approach to synthesizing images that are effective for training object detectors. Starting from a small set of real images, our algorithm estimates the rendering parameters required to synthesize similar images given a coarse 3D model of the target object. These parameters can then be reused to generate an unlimited number of training images of the object of interest in arbitrary 3D poses, which can then be used to increase classification performances. A key insight of our approach is that the synthetically generated images should be similar to real images, not in terms of image quality, but rather in terms of features used during the classifier training. We demonstrate the benefits of using such synthetically generated images in the context of drone detection, where limited amount of training data is available.
000199706 6531_ $$asynthetic image generation
000199706 6531_ $$aobject detection
000199706 700__ $$0246627$$g222094$$aRozantsev, Artem
000199706 700__ $$0240235$$g149007$$aLepetit, Vincent
000199706 700__ $$aFua, Pascal$$g112366$$0240252
000199706 8564_ $$uhttps://infoscience.epfl.ch/record/199706/files/optimisation_drone.avi$$s45163500
000199706 8564_ $$uhttps://infoscience.epfl.ch/record/199706/files/optimisation_plane.avi$$s44829338
000199706 8564_ $$uhttps://infoscience.epfl.ch/record/199706/files/report.pdf$$zn/a$$s29208999$$yn/a
000199706 8564_ $$uhttps://infoscience.epfl.ch/record/199706/files/supplementary_material.pdf$$s1544327
000199706 909C0 $$xU10659$$0252087$$pCVLAB
000199706 909CO $$qGLOBAL_SET$$pIC$$ooai:infoscience.tind.io:199706$$preport
000199706 917Z8 $$x222094
000199706 917Z8 $$x222094
000199706 917Z8 $$x222094
000199706 917Z8 $$x222094
000199706 937__ $$aEPFL-REPORT-199706
000199706 973__ $$aEPFL
000199706 980__ $$aREPORT