Correction formules d'erreurs et typos.

Problème d'erreurs constantes

FolderScraping.py

-import PIL
-from PIL import Image
-import pandas as pd
-import sklearn
-import os
-
-path = "./Init/Mer"
-img = Image.open("./Init/Mer/838s.jpg")
-print("Original size:", img.size)
-
-new_size = (img.width * 2, img.height * 2)  # Upscaling 2x
-resized_image = img.resize(new_size, Image.LANCZOS)  # Nearest-neighbor upscaling
-
-print("Resized size:", resized_image.size)
\ No newline at end of file

Main.py

-import PIL
-from PIL import Image
-import os
-import pandas as pd
-import numpy as np
-from sklearn.model_selection import train_test_split
-
-
-
-def computeHisto(image: PIL.Image.Image):
-    return image.histogram()
-
-def get_max_size(path1, path2):
-    max_size_x, max_size_y = 0, 0
-    max_size_x, max_size_y = compute_max_size_in_folder(max_size_x, max_size_y, path1)
-    max_size_x, max_size_y = compute_max_size_in_folder(max_size_x, max_size_y, path2)
-    return max_size_x, max_size_y
-
-
-def compute_max_size_in_folder(max_size_x, max_size_y, path1):
-    for image_path in os.listdir(path1):
-        full_path = os.path.join(path1, image_path)
-        image = Image.open(full_path)
-        if max_size_x * max_size_y < image.size[0] * image.size[1]:
-            max_size_x = image.size[0]
-            max_size_y = image.size[1]
-    return max_size_x, max_size_y
-
-
-def resizeImage (i, h, l) :
-    return i.resize((h,l), Image.LANCZOS)
-
-
-
-def buildSampleFromPath(path1, path2, size=0):
-    S = []
-
-    max_size = get_max_size(path1, path2)
-
-    for image_path in os.listdir(path1)[:size if size > 0 else -1]:
-
-        compute_dict(S, image_path, path1, 1, max_size)
-
-    for image_path in os.listdir(path2)[:size if size > 0 else -1]:
-        compute_dict(S, image_path, path2, -1, max_size)
-
-    return S
-
-
-
-def compute_dict(S, image_path, path, y_true_value, max_size: tuple):
-    full_path = os.path.join(path, image_path)
-    image = Image.open(full_path)
-    image = image.convert("RGB")
-    resized = resizeImage(image, *max_size)
-    #print("resized")
-    S.append({"name_path": full_path,
-              "resized_image": resized,
-              "X_histo": computeHisto(resized),
-              "y_true_class": y_true_value,
-              "y_predicted_class": None})
-
-
-
-def fitFromHisto(S, algo) :
-    df = pd.DataFrame(S)
-
-    #X = np.array(df["X_histo"])
-    X = np.array([np.array(l) for l in df["X_histo"]])
-    y = np.array(df["y_true_class"])
-
-
-    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
-
-    algo.fit(X_train, y_train)
-
-    return algo
-
-def predictFromHisto(S, model):
-    #for image in S:
-    #    image["y_predicted_class"] = model.predict(np.array(image["X_histo"]).reshape(1,-1))
-    tab = model.predict(np.array([x["X_histo"] for x in S]))
-    for i in range(len(S)):
-        S[i]["y_predicted_class"] = tab[i]
-
-def erreur_empirique(S):
-    error_count = 0
-    for image in S:
-        if image["y_true_class"] != image["y_predicted_class"]:
-            error_count += 1
-    return error_count/len(S)
-

TP.py

@@ -4,7 +4,6 @@ import os
 import pandas as pd
 import numpy as np
 from sklearn.model_selection import train_test_split
-from sklearn.naive_bayes import GaussianNB
 
 
 def buildSampleFromPath(path1, path2, size=0):
@@ -13,7 +12,7 @@ def buildSampleFromPath(path1, path2, size=0):
     used to train and test the model
     :param path1: path for the goods images, (score 1)
     :param path2: path for the bad images, (score -1)
-    :param size: Optionnal if you want to restrict the image pool
+    :param size: Optional if you want to restrict the image pool
     :return: list"""
     S = []
 
@@ -31,7 +30,7 @@ def buildSampleFromPath(path1, path2, size=0):
 def computeDict(image_path, path, y_true_value, max_size: tuple):
     """
     Middle function to construct each dict for each image. Resizing, and fetching the histogram,
-    by calling ohter functions
+    by calling other functions
     :param image_path: relative path of the image in the folder
     :param path: path of the folder containing the image
     :param y_true_value: is the image a good one (1) or a wrong one (-1)
@@ -149,7 +148,7 @@ def empiricalError(S):
     for image in S:
         if image["y_true_class"] != image["y_predicted_class"]:
             error_count += 1
-    return error_count / len(S)
+    return error_count
 
 def realError(S_test):
     """
@@ -161,6 +160,6 @@ def realError(S_test):
     for image in S_test:
         if image["y_predicted_class"] != image["y_true_class"]:
             error_count += 1
-    return error_count/len(S_test)
+    return error_count
 
 

TestMain.py → TestTP.py

-import os
 from PIL import Image
 from sklearn.naive_bayes import GaussianNB
 
@@ -29,14 +28,13 @@ def test_sample():
 
 def real_test():
     S = TP.buildSampleFromPath(path1_t, path2_t)
-    print("built")
-    classifieur, S_test, y_test = TP.fitFromHisto(S, GaussianNB())
-    TP.predictFromHisto(S, classifieur)
+    classifier, S_test, y_test = TP.fitFromHisto(S, GaussianNB())
+    TP.predictFromHisto(S, classifier)
     print(TP.empiricalError(S))
     print(TP.realError(S_test))
 
 if __name__ == "__main__":
-    #image = Image.open("/home/paul_pvc/PycharmProjects/pythonProject/TP_IA_L3/Init/Mer/838s.jpg")
+    #image = Image.open("./Init/Mer/838s.jpg")
     test_Histogram()
     #test_resizeImage(image, 1000, 1000)
     real_test()