Sparse iFRAME (Sparse Inhomogeneous Filters Random Field And Maximum Entropy)

Experiment 5: Detection via Sparse FRAME Model Learned by Generalized Gibbs Sampler and Generative Epsilon-boosting (with local normalization)

Code and dataset

(Optional: Matching Pursuit Version: Code)

Case 1: Tiger

(0) Training images

(1) Learning sparse-Frame model (by 6x6 multiple chains):

projected synthesized image (single chain)

projected synthesized image (multiple chains)

(2) Learning sequence

(3) Object detection by geometric transformation of template

(4) Cropped objects:

parameters setting:

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[ 1.4, 1, 0.7, 0.5]; DoGScaleList =[]; LocationShiftLimit=2; OrientShiftLimit=1; interval=5; #Wavelet=300; nIteration= (interval) x (#Wavelet) x 2; isLocalNormalize=true; isSeparate=false; localNormScaleFactor=2; thresholdFactor=0.01
Gibbs Sampler Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); 6x6 chains; sigsq=10; threshold_corrBB=0; lower_bound_rand = 0.001; upper_bound_rand = 0.999; c_val_list=-30:2:30
Detection Parameters: flipOrNot = false; rotateShiftLimit=4; numberResolution=15; scaleStepSize=0.1; initializedScaleFactor=170.

Case 2: Lion

(0) Training images

(1) Learning sparse-Frame model (by 6x6 multiple chains):

projected synthesized image (single chain)

projected synthesized image (multiple chains)

(2) Learning sequence

(3) Object detection by geometric transformation of template

(4) Cropped objects:

parameters setting:

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[ 1.4, 1, 0.7, 0.5]; DoGScaleList =[]; LocationShiftLimit=2; OrientShiftLimit=1; interval=5; #Wavelet=300; nIteration= (interval) x (#Wavelet) x 2; isLocalNormalize=true; isSeparate=false; localNormScaleFactor=2; thresholdFactor=0.01
Gibbs Sampler Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); 6x6 chains; sigsq=10; threshold_corrBB=0; lower_bound_rand = 0.001; upper_bound_rand = 0.999; c_val_list=-30:2:30
Detection Parameters: flipOrNot = false; rotateShiftLimit=4; numberResolution=15; scaleStepSize=0.1; initializedScaleFactor=150.

Case 3: Cat

(0) Training images

(1) Learning sparse-Frame model (by 6x6 multiple chains):

projected synthesized image (single chain)

projected synthesized image (multiple chains)

(2) Learning sequence

(3) Object detection by geometric transformation of template

(4) Cropped objects:

parameters setting:

General Parameters: nOrient = 16; sizeTemplatex=100; sizeTemplatey=100; GaborScaleList=[ 1.4, 1, 0.7, 0.5]; DoGScaleList =[]; LocationShiftLimit=2; OrientShiftLimit=1; interval=5; #Wavelet=300; nIteration= (interval) x (#Wavelet) x 2; isLocalNormalize=true; isSeparate=false; localNormScaleFactor=2; thresholdFactor=0.01
Gibbs Sampler Parameters: lambdaLearningRate = 0.1/sqrt(sigsq); 6x6 chains; sigsq=10; threshold_corrBB=0; lower_bound_rand = 0.001; upper_bound_rand = 0.999; c_val_list=-30:2:30
Detection Parameters: flipOrNot = false; rotateShiftLimit=4; numberResolution=15; scaleStepSize=0.1; initializedScaleFactor=200.