amirTamer commited on
Commit
8b4d0d5
·
1 Parent(s): 8fed8e8
Files changed (1) hide show
  1. app.py +38 -6
app.py CHANGED
@@ -3,6 +3,7 @@ import numpy as np
3
  from skimage.feature import hog
4
  import cv2
5
  import pickle
 
6
 
7
 
8
 
@@ -12,14 +13,35 @@ def load_pickle_file(file_name):
12
  with open(file_name, "rb") as f:
13
  loaded_file = pickle.load(f)
14
  return loaded_file
15
- except Exception:
 
16
  return None
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
17
 
18
  gnb_package = load_pickle_file("gnb_digit6_package.pkl")
19
  pca_multi = load_pickle_file("pcaMulti.pkl")
20
  rf_model = load_pickle_file("rf_hog_model.pkl")
21
- pca_multi = pca_multi['pca']
22
  pca_KNN = load_pickle_file("knn_pca.pkl")
 
 
23
 
24
  if gnb_package is not None:
25
  gnb_binary_model = gnb_package["model"]
@@ -53,12 +75,14 @@ svm_scaler = load_pickle_file("svmScaler.pkl")
53
 
54
 
55
  knn_model_binary = load_pickle_file('knn_final_model.pkl')
56
- knn_multi = load_pickle_file("knn_hog_pca.pkl")
57
 
58
  gnb_multimode_model = load_pickle_file('gnbMulti.pkl')
59
- gnb_multimode_model = gnb_multimode_model["model"]
 
60
  dt_multimode_model = load_pickle_file('dt_Multi.pkl')
61
- dt_multimode_model = dt_multimode_model['model']
 
62
  svm_multimode_model = None
63
  logreg_multimode_model = load_pickle_file('multiLog2_hog.pkl')
64
 
@@ -90,7 +114,7 @@ def preprocess_image(image, model_choice, mode_choice):
90
  return image_hog
91
  elif model_choice == "KNN":
92
  image_hog = image_hog.reshape(1, -1)
93
- image_pca = (image_hog - pca_KNN["pca_mean"]) @ pca_KNN["pca_components"].T
94
  return image_pca
95
 
96
 
@@ -153,12 +177,18 @@ def classify_knn(image):
153
 
154
 
155
  def classify_multimode_gnb(image):
 
 
 
156
  processed_image = preprocess_image(image, "GNB", "Multimode")
157
  prediction = gnb_multimode_model.predict(processed_image)
158
  return "GNB Multimode prediction: " + str(prediction[0])
159
 
160
 
161
  def classify_multimode_dt(image):
 
 
 
162
  processed_image = preprocess_image(image, "DT", "Multimode")
163
  prediction = dt_multimode_model.predict(processed_image)
164
  return "Decision Tree Multimode prediction: " + str(prediction[0])
@@ -183,6 +213,8 @@ def classify_multimode_logreg(image):
183
 
184
 
185
  def classify_multimode_knn(image):
 
 
186
 
187
  processed_image = preprocess_image(image, "KNN", "Multimode")
188
  prediction = knn_multi.predict(processed_image)
 
3
  from skimage.feature import hog
4
  import cv2
5
  import pickle
6
+ from KNN import CustomKNN
7
 
8
 
9
 
 
13
  with open(file_name, "rb") as f:
14
  loaded_file = pickle.load(f)
15
  return loaded_file
16
+ except pickle.UnpicklingError:
17
+ print(f"{file_name} is not a valid pickle. If it is tracked by Git LFS, run `git lfs pull`.")
18
  return None
19
+ except Exception as exc:
20
+ print(f"Could not load {file_name}: {exc}")
21
+ return None
22
+
23
+
24
+ def build_knn_from_pca_package(package):
25
+ if package is None:
26
+ return None
27
+
28
+ required_keys = {"best_k", "weights", "X_train", "y_train", "pca_mean", "pca_components"}
29
+ missing_keys = required_keys.difference(package)
30
+ if missing_keys:
31
+ print(f"knn_pca.pkl is missing keys: {sorted(missing_keys)}")
32
+ return None
33
+
34
+ return CustomKNN(k=package["best_k"], weights=package["weights"]).fit(
35
+ package["X_train"],
36
+ package["y_train"]
37
+ )
38
 
39
  gnb_package = load_pickle_file("gnb_digit6_package.pkl")
40
  pca_multi = load_pickle_file("pcaMulti.pkl")
41
  rf_model = load_pickle_file("rf_hog_model.pkl")
 
42
  pca_KNN = load_pickle_file("knn_pca.pkl")
43
+ if pca_multi is not None:
44
+ pca_multi = pca_multi['pca']
45
 
46
  if gnb_package is not None:
47
  gnb_binary_model = gnb_package["model"]
 
75
 
76
 
77
  knn_model_binary = load_pickle_file('knn_final_model.pkl')
78
+ knn_multi = build_knn_from_pca_package(pca_KNN)
79
 
80
  gnb_multimode_model = load_pickle_file('gnbMulti.pkl')
81
+ if gnb_multimode_model is not None:
82
+ gnb_multimode_model = gnb_multimode_model["model"]
83
  dt_multimode_model = load_pickle_file('dt_Multi.pkl')
84
+ if dt_multimode_model is not None:
85
+ dt_multimode_model = dt_multimode_model['model']
86
  svm_multimode_model = None
87
  logreg_multimode_model = load_pickle_file('multiLog2_hog.pkl')
88
 
 
114
  return image_hog
115
  elif model_choice == "KNN":
116
  image_hog = image_hog.reshape(1, -1)
117
+ image_pca = (image_hog - pca_KNN["pca_mean"]) @ pca_KNN["pca_components"].T
118
  return image_pca
119
 
120
 
 
177
 
178
 
179
  def classify_multimode_gnb(image):
180
+ if gnb_multimode_model is None or pca_multi is None:
181
+ return "GNB Multimode model could not be loaded. If this is a Git LFS file, run `git lfs pull` first."
182
+
183
  processed_image = preprocess_image(image, "GNB", "Multimode")
184
  prediction = gnb_multimode_model.predict(processed_image)
185
  return "GNB Multimode prediction: " + str(prediction[0])
186
 
187
 
188
  def classify_multimode_dt(image):
189
+ if dt_multimode_model is None:
190
+ return "Decision Tree Multimode model could not be loaded. If this is a Git LFS file, run `git lfs pull` first."
191
+
192
  processed_image = preprocess_image(image, "DT", "Multimode")
193
  prediction = dt_multimode_model.predict(processed_image)
194
  return "Decision Tree Multimode prediction: " + str(prediction[0])
 
213
 
214
 
215
  def classify_multimode_knn(image):
216
+ if knn_multi is None or pca_KNN is None:
217
+ return "KNN Multimode model could not be loaded. If this is a Git LFS file, run `git lfs pull` first."
218
 
219
  processed_image = preprocess_image(image, "KNN", "Multimode")
220
  prediction = knn_multi.predict(processed_image)