iFRAME (inhomogeneous Filters Random Field And Maximum Entropy)

Experiment 5.2: Unsupervised Learning of Codebooks (local normalization)

Code and dataset

Old results

Case 1: Grape

Parameters setting

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[0.7]; DoGScaleList=[]; sigsq = 10; locationShiftLimit = 3; orientShiftLimit = 1; numSketch = 30;  isGlobalNormalization = true; isLocalNormalize = true; resizeFactor = 1.2;
HMC Parameters:  lambdaLearningRate = 0.1/sqrt(sigsq); epsilon = 0.03; L = 10; nIteraton =40; 12x12 chains; 
Coodebook Parameters: rotateShiftLimit = 16; allResolution = [0.8, 1, 1.2]; #EM iteration = 20; numCluster = 2; maxNumClusterMember = 1000; locationPerturbationFraction = 0.4; SUM2mapBoundaryFraction=2.5;

Case 2: Beehive

training image

testing image

Parameters setting

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[0.7]; DoGScaleList=[]; sigsq = 10; locationShiftLimit = 3; orientShiftLimit = 1; numSketch = 30;  isGlobalNormalization = true; isLocalNormalize = true; resizeFactor = 1.2;
HMC Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); epsilon = 0.03; L = 10; nIteraton =40; 12x12 chains; 
Coodebook Parameters: rotateShiftLimit = 16; allResolution = [0.8, 1, 1.2]; #EM iteration = 15; numCluster = 1; maxNumClusterMember = 1000; locationPerturbationFraction = 0.4; SUM2mapBoundaryFraction=4;

Case 3: Peanut

training image

testing image

Parameters setting

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[0.7]; DoGScaleList=[]; sigsq = 10; locationShiftLimit = 3; orientShiftLimit = 1; numSketch = 30;  isGlobalNormalization = true; isLocalNormalize = true; resizeFactor = 1.4;
HMC Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); epsilon = 0.03; L = 10; nIteraton =40; 12x12 chains; 
Coodebook Parameters: rotateShiftLimit = 16; allResolution = [0.8, 1, 1.2]; #EM iteration = 7; numCluster = 1; maxNumClusterMember = 1000; locationPerturbationFraction = 0.4; SUM2mapBoundaryFraction=4;

Case 4: Stone wall

Parameters setting

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[0.7]; DoGScaleList=[]; sigsq = 10; locationShiftLimit = 3; orientShiftLimit = 1; numSketch = 32;  isGlobalNormalization = true; isLocalNormalize = true; resizeFactor = 1.4;
HMC Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); epsilon = 0.03; L = 10; nIteraton =40; 12x12 chains; 
Coodebook Parameters: rotateShiftLimit = 16; allResolution = [0.8, 1, 1.2]; #EM iteration = 15; numCluster = 2; maxNumClusterMember = 1000; locationPerturbationFraction = 0.4; SUM2mapBoundaryFraction=2.5;

Case 5: Cat

training images

testing image

Parameters setting

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[0.7]; DoGScaleList=[]; sigsq = 10; locationShiftLimit = 3; orientShiftLimit = 1; numSketch = 50;  isGlobalNormalization = true; isLocalNormalize = true; resizeFactor = [160 160];
HMC Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); epsilon = 0.03; L = 10; nIteraton =40; 12x12 chains; 
Coodebook Parameters: rotateShiftLimit = 2; allResolution = [0.8, 1, 1.2]; #EM iteration = 15; numCluster =8; maxNumClusterMember = 1000; locationPerturbationFraction = 0.4; SUM2mapBoundaryFraction=2.3;

Case 6: Lotus

Parameters setting

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[0.7]; DoGScaleList=[]; sigsq = 10; locationShiftLimit = 3; orientShiftLimit = 1; numSketch = 35;  isGlobalNormalization = true; isLocalNormalize = true; resizeFactor =1;
HMC Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); epsilon = 0.03; L = 10; nIteraton =40; 12x12 chains; 
Coodebook Parameters: rotateShiftLimit = 16; allResolution = [0.8, 1, 1.2]; #EM iteration = 15; numCluster =2; maxNumClusterMember = 1000; locationPerturbationFraction = 0.4; SUM2mapBoundaryFraction=3.9;