MMU2 optimal parameters

python/calculate_dist_mat_mmu2.py

@@ -0,0 +1,117 @@
+#------------------------------------------------------------------------------
+#	Libraries
+#------------------------------------------------------------------------------
+import os
+import numpy as np
+from glob import glob
+from time import time
+from random import shuffle
+from matplotlib import pyplot as plt
+from itertools import repeat
+from multiprocessing import Pool, cpu_count
+
+from fnc.extractFeature import extractFeature
+from fnc.matching import calHammingDist
+
+
+#------------------------------------------------------------------------------
+#   Parameters
+#------------------------------------------------------------------------------
+CASIA1_DIR = "/home/antiaegis/Downloads/Iris-Recognition/MMU2"
+EYELASHES_THRES = 10
+N_IMAGES = 5
+
+
+#------------------------------------------------------------------------------
+#   Pool function of extracting feature
+#------------------------------------------------------------------------------
+def pool_func_extract_feature(args):
+	im_filename, eyelashes_thres, use_multiprocess = args
+
+	template, mask, im_filename = extractFeature(
+		im_filename=im_filename,
+		eyelashes_thres=eyelashes_thres,
+		use_multiprocess=use_multiprocess,
+	)
+	return template, mask, im_filename
+
+
+#------------------------------------------------------------------------------
+#   Pool function of calculating Hamming distance
+#------------------------------------------------------------------------------
+def pool_func_calHammingDist(args):
+	template1, mask1, template2, mask2 = args
+	dist = calHammingDist(template1, mask1, template2, mask2)
+	return dist
+
+
+#------------------------------------------------------------------------------
+#	Main execution
+#------------------------------------------------------------------------------
+# Get identities of MMU2 dataset
+identities = glob(os.path.join(CASIA1_DIR, "**"))
+identities = sorted([os.path.basename(identity) for identity in identities])
+n_identities = len(identities)
+print("Number of identities:", n_identities)
+
+
+# Construct a dictionary of files
+files_dict = {}
+image_files = []
+for identity in identities:
+	if identity=="50":
+		continue
+
+	files = glob(os.path.join(CASIA1_DIR, identity, "*.*"))
+	shuffle(files)
+	files_dict[identity] = files[:N_IMAGES]
+	# print("Identity %s: %d images" % (identity, len(files_dict[identity])))
+	image_files += files[:N_IMAGES]
+
+n_image_files = len(image_files)
+print("Number of image files:", n_image_files)
+
+
+# Extract features
+args = zip(image_files, repeat(EYELASHES_THRES), repeat(False))
+pools = Pool(processes=cpu_count())
+
+start_time = time()
+features = list(pools.map(pool_func_extract_feature, args))
+finish_time = time()
+print("Extraction time: %.3f [s]" % (finish_time-start_time))
+
+
+# Calculate the distances
+args = []
+for i in range(n_image_files):
+	for j in range(n_image_files):
+		if i>=j:
+			continue
+
+		arg = (features[i][0], features[i][1], features[j][0], features[j][1])
+		args.append(arg)
+print("Number of pairs:", len(args))
+
+start_time = time()
+distances = pools.map(pool_func_calHammingDist, args)
+finish_time = time()
+print("Extraction time: %.3f [s]" % (finish_time-start_time))
+
+
+# Construct a distance matrix
+dist_mat = np.zeros([n_image_files, n_image_files])
+k = 0
+for i in range(n_image_files):
+	for j in range(n_image_files):
+		if i<j:
+			dist_mat[i, j] = distances[k]
+			k += 1
+		elif i>j:
+			dist_mat[i, j] = dist_mat[j, i]
+
+np.save("dist_mat_mmu2.npy", dist_mat)
+
+plt.figure()
+plt.imshow(dist_mat)
+plt.show()

python/eval_thres_mmu2.py

@@ -0,0 +1,49 @@
+#------------------------------------------------------------------------------
+#	Libraries
+#------------------------------------------------------------------------------
+import numpy as np
+from matplotlib import pyplot as plt
+
+
+#------------------------------------------------------------------------------
+#   Parameters
+#------------------------------------------------------------------------------
+DIST_MAT_FILE = "dist_mat_mmu2.npy"
+THRESHOLDS = np.linspace(start=0.0, stop=1.0, num=100)
+NUM_IMAGES = 5
+
+
+#------------------------------------------------------------------------------
+#   Main execution
+#------------------------------------------------------------------------------
+dist_mat = np.load(DIST_MAT_FILE)
+
+ground_truth = np.zeros_like(dist_mat, dtype=int)
+for i in range(ground_truth.shape[0]):
+    for j in range(ground_truth.shape[1]):
+        if i//NUM_IMAGES == j//NUM_IMAGES:
+            ground_truth[i, j] = 1
+
+accuracies, precisions, recalls, fscores = [], [], [], []
+for threshold in THRESHOLDS:
+    decision_map = (dist_mat<=threshold).astype(int)
+    accuracy = (decision_map==ground_truth).sum() / ground_truth.size
+    precision = (ground_truth*decision_map).sum() / decision_map.sum()
+    recall = (ground_truth*decision_map).sum() / ground_truth.sum()
+    fscore = 2*precision*recall / (precision+recall)
+    accuracies.append(accuracy)
+    precisions.append(precision)
+    recalls.append(recall)
+    fscores.append(fscore)
+
+print("Max fscore:", max(fscores))
+print("Best threshold:", THRESHOLDS[fscores.index(max(fscores))])
+
+plt.figure()
+plt.plot(THRESHOLDS, accuracies, "-or")
+plt.plot(THRESHOLDS, precisions, "-vb")
+plt.plot(THRESHOLDS, recalls, "-*g")
+plt.plot(THRESHOLDS, fscores, "-sc")
+plt.legend(["accuracy", "precision", "recall", "fscore"])
+plt.grid(True)
+plt.show()