- 02 Dec, 2013 2 commits
- 
- 
Davis King authoredrotate_image_dataset(). 
- 
davis@potato.localnet authored
 
- 
- 30 Nov, 2013 6 commits
- 
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authoredthe 2011 paper Introduction to dual decomposition for inference by David Sontag, Amir Globerson, and Tommi Jaakkola since the original 2008 paper had an error in the algorithm that negatively effected its convergence. Thanks to James Gunning for pointing this out. 
 
- 
- 28 Nov, 2013 1 commit
- 
- 
Davis King authored
 
- 
- 24 Nov, 2013 8 commits
- 
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authoredmade during object detection. 
- 
Davis King authored
- 
Davis King authoredchanged the interfaces to these functions slightly. Instead of taking a double that determines how we decide if boxes match, they now take a test_box_overlap object. 
- 
Davis King authored
- 
Davis King authoredAlso added overlaps_any_box(). 
 
- 
- 23 Nov, 2013 3 commits
- 
- 
Davis King authoreddefaults to using exactly the PASCAL VOC match criterion. 
- 
Davis King authored
- 
Davis King authored
 
- 
- 22 Nov, 2013 7 commits
- 
- 
Davis King authoredrectangles instead of just one. 
- 
Davis King authoredin each image. 
- 
Davis King authoredrectangles. 
- 
Davis King authoredto/from the xml files. 
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
 
- 
- 21 Nov, 2013 1 commit
- 
- 
Davis King authoredmultiple object detectors. 
 
- 
- 20 Nov, 2013 1 commit
- 
- 
Davis King authored
 
- 
- 19 Nov, 2013 2 commits
- 
- 
Davis King authored
- 
Davis King authoredthe fhog_filterbank each time the detector executes. 
 
- 
- 18 Nov, 2013 9 commits
- 
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authored
- 
Davis King authoredso they can run each binary trainer on a different core. 
- 
Davis King authoredby reference since this causes problems if the probabilistic adapter is saved for later use. Now it will simply hold the trainer by value. 
- 
Davis King authored
 
-