a, q, pos, cnt = accuracy.accuracy_matches(q_path, imagematches, 20)
print('Accuracy =', a, '%', '| Quality:', q)
print('Count', cnt, ' | position', pos)
#########--------------------------------SIFT
import ImageSearch_Algo_SIFT
#generation
# Hyper-Parameters for SIFT comparison
sift_features_limit = 100
lowe_ratio = 0.75
predictions_count = 50
IMGDIR = "./imagesbooks/"
imagepaths = list(paths.list_images(IMGDIR))
mydatasift, mytimesift = ImageSearch_Algo_SIFT.gen_sift_features(
imagepaths, 200)
print('generationtime', mytimesift)
savefile = 'data/SIFT_features_519set_kp200_pandas'
ImageSearch_Algo_SIFT.SIFT_SAVE_FEATURES(mydatasift, savefile)
# mydataSIFT = ImageSearch_Algo_SIFT.SIFT_LOAD_FEATURES(savefile)
# q_paths = random.sample(imagepaths, 50) # random sample 100 items in list
# q_paths = ['./imagesbooks/ukbench05960.jpg', './imagesbooks/ukbench00459.jpg', './imagesbooks/ukbench06010.jpg', './imagesbooks/ukbench06104.jpg', './imagesbooks/ukbench00458.jpg']
q_paths = [
'./imagesbooks/ukbench05960.jpg', './imagesbooks/ukbench00459.jpg',
'./imagesbooks/ukbench06010.jpg', './imagesbooks/ukbench06104.jpg',
'./imagesbooks/ukbench00458.jpg', './imagesbooks/ukbench00248.jpg',
'./imagesbooks/ukbench06408.jpg', './imagesbooks/ukbench00303.jpg',
'./imagesbooks/ukbench03124.jpg', './imagesbooks/ukbench05776.jpg',
# --------------- IMAGES ----------------------#
imagepaths = sorted (list(paths.list_images(IMGDIR)))
myDataFiles = pd.DataFrame( {'file' : imagepaths })
# ----------- GENERATE ALL FEATURES & SAVE ------------ #
# GEN SIFT
sift_features_limit = SIFT_FEATURES_LIMIT
lowe_ratio = LOWE_RATIO
predictions_count = SIFT_PREDICTIONS_COUNT
mydataSIFT, mytime1 = ImageSearch_Algo_SIFT.gen_sift_features(
imagepaths, sift_features_limit)
print("SIFT Feature Generation time :", mytime1)
savefile = 'data/' + TESTNAME + '_PandasDF_SIFT_Features_kp'+ str(sift_features_limit)
ImageSearch_Algo_SIFT.SIFT_SAVE_FEATURES (mydataSIFT, savefile)
print("SIFT Feature saved to : ", savefile)
# -- END
# GEN ORB
orb_features_limit = ORB_FEATURES_LIMIT
mydataORB, mytime1 = ImageSearch_Algo_ORB.GEN_ORB_FEATURES(imagepaths, orb_features_limit)
print("ORB Feature Generation time :", mytime1)
savefile = 'data/' + TESTNAME + '_PandasDF_ORB_Features_kp'+ str(orb_features_limit)
ImageSearch_Algo_ORB.ORB_SAVE_FEATURES (mydataORB, savefile)
print("ORB Feature saved to : ", savefile)
# -- END
def GENERATE(TESTNAME, IMGDIR):
'''Sequesnce of tests
TESTNAME : name of the test (suffix to all the pickles, csv dumped)
IMGDIR : DIRECTORY PATH CONTAINING IMAGES
'''
# --------------- CONFIG PARAMETERS ----------------------#
ORB_FEATURES_LIMIT = 100
ORB_FEATURES_LIMIT2 = 500
ORB_N_CLUSTERS = 500
ORB_N_CLUSTERS2 = 50 # 500 # (option2) #optimal = 50 for 500 features
SIFT_FEATURES_LIMIT = 100
SIFT_FEATURES_LIMIT2 = 300
SIFT_N_CLUSTERS = 500
SIFT_N_CLUSTERS2 = 50 # 100 # (option2) #optimal = 50 for 300 features
LOWE_RATIO = 0.7
SIFT_PREDICTIONS_COUNT = 100
RGB_PARAMETERCORRELATIONTHRESHOLD = 0.70 # not needed for generation
kneeHSV = 2
kneeRGB = 2
kneeORB = 2
kneeSIFT = 2
HASHLENGTH = 16
# --------------- IMAGES ----------------------#
imagepaths = (list(paths.list_images(IMGDIR)))
myDataFiles = pd.DataFrame({'file': imagepaths})
# ----------- GENERATE ALL FEATURES & SAVE ------------ #
print('Testname: ', TESTNAME)
print("Generating Features for ", len(imagepaths), "images in", IMGDIR)
# GEN SIFT
sift_features_limit = SIFT_FEATURES_LIMIT
lowe_ratio = LOWE_RATIO
predictions_count = SIFT_PREDICTIONS_COUNT
mydataSIFT, mytime1 = ImageSearch_Algo_SIFT.gen_sift_features(
imagepaths, sift_features_limit)
print("SIFT Feature Generation time :", mytime1)
savefile = 'data/' + TESTNAME + '_PandasDF_SIFT_Features_kp' + str(
sift_features_limit)
ImageSearch_Algo_SIFT.SIFT_SAVE_FEATURES(mydataSIFT, savefile)
print("SIFT Feature saved to : ", savefile)
# -- END
# GEN SIFT 2 # for TREE kp=300, n_cluster=50
sift_features_limit = SIFT_FEATURES_LIMIT2
lowe_ratio = LOWE_RATIO
predictions_count = SIFT_PREDICTIONS_COUNT
mydataSIFT2, mytime1 = ImageSearch_Algo_SIFT.gen_sift_features(
imagepaths, sift_features_limit)
print("SIFT Feature Generation time :", mytime1)
savefile = 'data/' + TESTNAME + '_PandasDF_SIFT_Features_kp' + str(
sift_features_limit)
ImageSearch_Algo_SIFT.SIFT_SAVE_FEATURES(mydataSIFT2, savefile)
print("SIFT Feature saved to : ", savefile)
# -- END
# GEN ORB
orb_features_limit = ORB_FEATURES_LIMIT
mydataORB, mytime1 = ImageSearch_Algo_ORB.GEN_ORB_FEATURES(
imagepaths, orb_features_limit)
print("ORB Feature Generation time :", mytime1)
savefile = 'data/' + TESTNAME + '_PandasDF_ORB_Features_kp' + str(
orb_features_limit)
ImageSearch_Algo_ORB.ORB_SAVE_FEATURES(mydataORB, savefile)
print("ORB Feature saved to : ", savefile)
# -- END
# GEN ORB 2 # for TREE kp=500, n_cluster=50
orb_features_limit = ORB_FEATURES_LIMIT2
mydataORB2, mytime1 = ImageSearch_Algo_ORB.GEN_ORB_FEATURES(
imagepaths, orb_features_limit)
print("ORB Feature Generation time :", mytime1)
savefile = 'data/' + TESTNAME + '_PandasDF_ORB_Features_kp' + str(
orb_features_limit)
ImageSearch_Algo_ORB.ORB_SAVE_FEATURES(mydataORB2, savefile)
print("ORB Feature saved to : ", savefile)
# -- END
# GEN RGB
parametercorrelationthreshold = 0.70 # not needed for generation
mydataRGB, mytime = ImageSearch_Algo_RGB.RGB_GEN(imagepaths)
print('RGB Feature Generation time', mytime)
savefile = 'data/' + TESTNAME + '_PandasDF_RGB_Features'
ImageSearch_Algo_RGB.RGB_SAVE_FEATURES(mydataRGB, savefile)
print("RGB Feature saved to : ", savefile)
# -- END
# GEN HSV
mydataHSV, mytime = ImageSearch_Algo_HSV.HSV_GEN(imagepaths)
print('HSV Feature Generation time', mytime)
savefile = 'data/' + TESTNAME + '_PandasDF_HSV_Features'
ImageSearch_Algo_HSV.HSV_SAVE_FEATURES(mydataHSV, savefile)
print("HSV Feature saved to : ", savefile)
# -- END
# GEN HASH
mydataHASH, mytime = ImageSearch_Algo_Hash.HASH_GEN(imagepaths, HASHLENGTH)
print("HASH Features Generation time :", mytime)
savefile = 'data/' + TESTNAME + '_PandasDF_HASH_Features'
ImageSearch_Algo_Hash.HASH_SAVE_FEATURES(mydataHASH, savefile)
# -- END
print("## Feature Generation Complete.")
# ----------- GENERATE ALL TREES ------------ #
# RGB TREE
savefile = 'data/' + TESTNAME + '_RGB_Tree'
myRGBtree = ImageSearch_Algo_RGB.RGB_Create_Tree(mydataRGB,
savefile=savefile)
# HSV TREE
savefile = 'data/' + TESTNAME + '_HSV_Tree'
myHSVtree = ImageSearch_Algo_HSV.HSV_Create_Tree(mydataHSV,
savefile=savefile)
# HASH TREE
AlgoGenList = ['whash', 'phash', 'dhash', 'ahash']
for algo in AlgoGenList:
savefile = 'data/' + TESTNAME + '_HASH_Tree_' + str(algo)
myHASHTree = ImageSearch_Algo_Hash.HASH_Create_Tree(mydataHASH,
savefile=savefile,
hashAlgo=algo)
# HASH TREE USE HYBRID HASH
HybridAlgoList = ['whash', 'ahash']
savefile = 'data/' + TESTNAME + '_HASH_Hybrid_Tree_' + str(
('_').join(HybridAlgoList))
myHybridtree = ImageSearch_Algo_Hash.HASH_CREATE_HYBRIDTREE(
mydataHASH, savefile, HybridAlgoList)
# SIFT FV Tree and Cluster
n_clusters = SIFT_N_CLUSTERS
savefile = 'data/' + TESTNAME + '_SIFT_Tree_Cluster' + str(n_clusters)
mySIFTtree, mySIFTmodel, mySIFTFVHist = ImageSearch_Algo_SIFT.SIFT_CREATE_TREE_MODEL(
mydataSIFT, savefile, n_clusters)
# SIFT FV Tree and Cluster with kp=300, n_cluster=50
n_clusters = SIFT_N_CLUSTERS2
savefile = 'data/' + TESTNAME + '_SIFT_Tree_Cluster' + str(
n_clusters) + 'kp' + str(SIFT_FEATURES_LIMIT2)
mySIFTtree2, mySIFTmodel2, mySIFTFVHist2 = ImageSearch_Algo_SIFT.SIFT_CREATE_TREE_MODEL(
mydataSIFT2, savefile, n_clusters)
# ORB FV Tree and Cluster
n_clusters = ORB_N_CLUSTERS
savefile = 'data/' + TESTNAME + '_ORB_Tree_Cluster' + str(n_clusters)
myORBtree, myORBmodel, myORBFVHist = ImageSearch_Algo_ORB.ORB_CREATE_TREE_MODEL(
mydataORB, savefile, n_clusters)
# ORB FV Tree and Cluster with kp=500, n_cluster=50
n_clusters = ORB_N_CLUSTERS2
savefile = 'data/' + TESTNAME + '_ORB_Tree_Cluster' + str(
n_clusters) + 'kp' + str(ORB_FEATURES_LIMIT2)
myORBtree2, myORBmodel2, myORBFVHist2 = ImageSearch_Algo_ORB.ORB_CREATE_TREE_MODEL(
mydataORB2, savefile, n_clusters)
print("## Tree Generation Complete.")