[Grayscale] Boundary Detection Benchmark: Algorithm "Texture Gradient"

This algorithm uses local texture gradients to detect boundaries. A logistic is used to model the posterior probability of a boundary. The texture gradient is computed as the chi-squared difference in the distribution of textons in two half discs centered at a pixel and divided in half at the putative boundary orientation. Textons are computed by clustering filterbank responses using k-means, so that they model the joint distribution of filter responses. The filters are standard even- and odd-symmetric quadrature pair elongated linear filters. See our NIPS and PAMI papers in the Grouping area of the Berkeley Vision Group web pages for additional details.

Click on an image for additional details.

#1 (119082) F=0.63

#2 (170057) F=0.62

#3 (58060) F=0.49

#4 (163085) F=0.48

#5 (42049) F=0.86

#6 (167062) F=0.64

#7 (157055) F=0.65

#8 (295087) F=0.73

#9 (24077) F=0.66

#10 (78004) F=0.70

#11 (220075) F=0.58

#12 (45096) F=0.68

#13 (38092) F=0.67

#14 (43074) F=0.53

#15 (16077) F=0.55

#16 (86000) F=0.62

#17 (101085) F=0.67

#18 (219090) F=0.71

#19 (89072) F=0.62

#20 (300091) F=0.60

#21 (126007) F=0.66

#22 (156065) F=0.55

#23 (76053) F=0.51

#24 (296007) F=0.58

#25 (175032) F=0.49

#26 (253027) F=0.70

#27 (304034) F=0.37

#28 (86016) F=0.47

#29 (103070) F=0.63

#30 (8023) F=0.37

#31 (260058) F=0.57

#32 (41033) F=0.52

#33 (291000) F=0.52

#34 (109053) F=0.58

#35 (130026) F=0.46

#36 (241004) F=0.72

#37 (108082) F=0.47

#38 (285079) F=0.66

#39 (147091) F=0.65

#40 (69040) F=0.46

#41 (14037) F=0.58

#42 (54082) F=0.57

#43 (189080) F=0.70

#44 (229036) F=0.60

#45 (62096) F=0.73

#46 (271035) F=0.62

#47 (167083) F=0.62

#48 (12084) F=0.43

#49 (69015) F=0.69

#50 (148089) F=0.62

#51 (160068) F=0.83

#52 (145086) F=0.62

#53 (216081) F=0.69

#54 (97033) F=0.62

#55 (182053) F=0.64

#56 (208001) F=0.65

#57 (19021) F=0.58

#58 (227092) F=0.71

#59 (134035) F=0.68

#60 (223061) F=0.71

#61 (253055) F=0.63

#62 (148026) F=0.53

#63 (210088) F=0.66

#64 (86068) F=0.48

#65 (3096) F=0.88

#66 (41069) F=0.60

#67 (21077) F=0.64

#68 (196073) F=0.69

#69 (108070) F=0.44

#70 (123074) F=0.55

#71 (376043) F=0.51

#72 (306005) F=0.65

#73 (38082) F=0.52

#74 (33039) F=0.52

#75 (108005) F=0.54

#76 (106024) F=0.66

#77 (302008) F=0.70

#78 (102061) F=0.51

#79 (197017) F=0.70

#80 (299086) F=0.68

#81 (37073) F=0.65

#82 (241048) F=0.63

#83 (65033) F=0.63

#84 (55073) F=0.49

#85 (66053) F=0.65

#86 (143090) F=0.57

#87 (85048) F=0.65

#88 (42012) F=0.53

#89 (351093) F=0.63

#90 (361010) F=0.70

#91 (175043) F=0.47

#92 (87046) F=0.59

#93 (105025) F=0.61

#94 (236037) F=0.47

#95 (101087) F=0.69

#96 (304074) F=0.59

#97 (296059) F=0.80

#98 (159008) F=0.54

#99 (385039) F=0.70

#100 (69020) F=0.56

Page generated on 20-Feb-2013 11:08:21.