fix: exports of the datasets

src/export_dataset.py

@@ -58,7 +58,7 @@ def export_embeddings():
         # write the header
         writer.writerow(header)
 
-        seamese = models.load_model(get_modeldir('seamese1.tf'))
+        seamese = models.load_model(get_modeldir('seamese_cifar10.tf'))
 
         embedding_vds = (cifar10_vds.map(process_images_couple).batch(batch_size=32, drop_remainder=False))
         print('predicting embeddings')
@@ -66,12 +66,12 @@ def export_embeddings():
         print('embeddings done')
 
         for i, (label) in enumerate(cifar10_labels):
-            label_str = label
+            label_str = ','.join(map(str, label))
             value_str = ','.join(map(str, embeddings[i]))
             writer.writerow([i, label_str, value_str])
 
 
 # export_hsv()
-export_sift()
-# export_embeddings()
+# export_sift()
+export_embeddings()
 print('done')

src/images.py

@@ -9,7 +9,9 @@ from tensorflow.keras import datasets
 
 # Load dataset
 (train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()
-cifar10_vds = tf.data.Dataset.from_tensor_slices((np.concatenate([train_images, test_images]), np.concatenate([train_labels, test_labels])))
+cifar10_images = np.concatenate([train_images, test_images])
+cifar10_labels = np.concatenate([train_labels, test_labels])
+cifar10_vds = tf.data.Dataset.from_tensor_slices((cifar10_images, cifar10_labels))
 
 # test HSV
 print('test HSV')
@@ -18,4 +20,14 @@ plot_hsv(cifar10_vds)
 print('test SIFT')
 plot_sift(cifar10_vds)
 
+# 906, 1692, 1711, 2610, 3259, 3418, 3789, 4277, 4975, 5010, 5255, 5867, 5988, 6406, 7089, 7365, 8072
+# 8443, 8998, 9008, 9323, 9664, 9881, 9903, 9985, 10095, 11650, 13043, 13075, 13841, 14698, 15443
+# 16004, 16733, 16888, 18948, 19378, 20015, 20233, 20467, 20621, 20696, 20778, 22672, 22804, 22904
+# 23252, 23654, 23985, 25236, 25734, 25931, 27596, 27931, 28016, 28300, 28387, 28807, 30029, 31581
+# 32024, 32117, 32629, 32861, 33328, 33489, 33589, 34466, 35063, 35202, 35719, 35877, 35985, 36560
+# 36777, 37358, 37439, 38224, 38345, 39942, 40389, 40621, 40864, 41454, 41902, 42017, 43593, 44207
+# 44226, 44257, 45801, 47091, 47375, 48663, 48690, 48884, 52366, 52622, 52847, 53227, 53248, 53423
+# 53429, 53444, 53660, 53759, 53952, 54957, 55164, 55189, 55762, 56549, 56574, 57105, 57171, 58485
+# 58572, 58826, 59318, 59970
+
 print('done')

src/siamese.py

@@ -79,4 +79,4 @@ embedding = alexnet(im_input)
 image_vector = projection_model(embedding)
 inference_model = Model(inputs=im_input, outputs=image_vector)
 
-inference_model.save(get_modeldir('seamese1.tf'), save_format='tf', include_optimizer=False)
+inference_model.save(get_modeldir('seamese_cifar10.tf'), save_format='tf', include_optimizer=False)

src/utils/hsv.py

@@ -2,7 +2,7 @@ import matplotlib.pyplot as plt
 import numpy as np
 import cv2
 
-from src.utils.common import subplot_image
+from src.utils.common import *
 
 
 def extract_hsv(image):
@@ -10,14 +10,15 @@ def extract_hsv(image):
     hsv = cv2.cvtColor(image, cv2.COLOR_RGB2HSV)
     # The ranges of the 3 HSV channels in opencv are 0-180, 0-256, 0-256 respectively
     # Bins is set to 1365, so that each picture can be represented by a 4000-dimensional vector
-    histh = cv2.calcHist([hsv], [0], None, [1365], [0, 180])
-    hists = cv2.calcHist([hsv], [1], None, [1365], [0, 256])
-    histv = cv2.calcHist([hsv], [2], None, [1365], [0, 256])
+    histh = cv2.calcHist([hsv], [0], None, [170], [0, 180])
+    hists = cv2.calcHist([hsv], [1], None, [171], [0, 256])
+    histv = cv2.calcHist([hsv], [2], None, [171], [0, 256])
     # normalize the histogram
     histh /= histh.sum()
     hists /= hists.sum()
     histv /= histv.sum()
-    hist_array = np.array([histh, hists, histv])
+    hist_array = np.append(histh, hists)
+    hist_array = np.append(hist_array, histv)
     # return the flattened histogram as the feature vector
     return histh, hists, histv, hist_array.flatten()
 
@@ -25,9 +26,12 @@ def extract_hsv(image):
 def plot_hsv(dataset):
     plt.figure(figsize=(20, 20))
     for i, (image, label) in enumerate(dataset.take(3)):
-        subplot_image(3, 2, i * 2 + 1, image, "Original image")
+        # from smaller image only smaller number of key points can be extracted
+        img = cv2.resize(image.numpy(), target_shape)
 
-        hist0_s, hist1_s, hist2_s, hist_s = extract_hsv(image.numpy())
+        subplot_image(3, 2, i * 2 + 1, img, "Original image")
+
+        hist0_s, hist1_s, hist2_s, hist_s = extract_hsv(img)
         # print('the length of histogram of the sample', len(hist_s))
 
         # subplot_image(3, 2, i * 2 + 2, image, "HSV Histogram")

src/utils/sift.py

 import matplotlib.pyplot as plt
 import cv2
 
-from src.utils.common import subplot_image
+from src.utils.common import *
 
 
 def extract_sift(image):
-    sift = cv2.SIFT_create(32)
+    sift = cv2.SIFT_create(8)
     # Calculate the keypoint and each point description of the image
     keypoints, features = sift.detectAndCompute(image, None)
     return keypoints, features
@@ -14,10 +14,10 @@ def extract_sift(image):
 def plot_sift(dataset):
     plt.figure(figsize=(20, 20))
     for i, (image, label) in enumerate(dataset.take(3)):
-        subplot_image(3, 2, i * 2 + 1, image, "Original image")
-
         # from smaller image only smaller number of key points can be extracted
-        img = cv2.resize(image.numpy(), (230, 230))
+        img = cv2.resize(image.numpy(), target_shape)
+
+        subplot_image(3, 2, i * 2 + 1, img, "Original image")
 
         keypoints, features = extract_sift(img)
         img_kp = img.copy()