000085972 001__ 85972
000085972 005__ 20190316233737.0
000085972 037__ $$aREP_WORK
000085972 245__ $$aIntegrating co-occurrence and spatial contexts on patch-based scene segmentation
000085972 269__ $$a2005
000085972 260__ $$bIDIAP$$c2005
000085972 336__ $$aReports
000085972 500__ $$aPublished in Beyond Patches Workshop, in conjuction with CVPR 2006
000085972 520__ $$aWe present a novel approach for contextual segmentation of complex visual scenes, based on the use of bags of local invariant features (visterms) and probabilistic aspect models. Our approach uses context in two ways: (1) by using the fact that specific learned aspects correlate with the semantic classes, which resolves some cases of visual polysemy, and (2) by formalizing the notion that scene context is image-specific -what an individual visterm represents depends on what the rest of the visterms in the same bag represent too-. We demonstrate the validity of our approach on a man-made vs. natural visterm classification problem. Experiments on an image collection of complex scenes show that the approach improves region discrimination, producing satisfactory results, and outperforming a non-contextual method. Furthermore, through the later use of a Markov Random Field model, we also show that co-occurrence and spatial contextual information can be conveniently integrated for improved visterm classification.
000085972 6531_ $$avision
000085972 700__ $$aMonay, Florent
000085972 700__ $$aQuelhas, Pedro
000085972 700__ $$0243995$$aOdobez, Jean-Marc$$g161663
000085972 700__ $$0241066$$aGatica-Perez, Daniel$$g171600
000085972 8564_ $$uhttp://publications.idiap.ch/downloads/reports/2005/monay-idiap-rr-05-30.pdf$$zURL
000085972 8564_ $$s1082180$$uhttps://infoscience.epfl.ch/record/85972/files/monay-idiap-rr-05-30.pdf$$zn/a
000085972 909C0 $$0252189$$pLIDIAP$$xU10381
000085972 909CO $$ooai:infoscience.tind.io:85972$$pSTI$$preport$$qGLOBAL_SET
000085972 937__ $$aEPFL-REPORT-85972
000085972 970__ $$amonay:rr05-30/LIDIAP
000085972 973__ $$aEPFL$$sPUBLISHED
000085972 980__ $$aREPORT