Predictions

predict.py

+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+
+@author: Gunasinghe M.D.
+"""
+
+import pandas as pd
+import numpy as np
+import cv2
+import argparse
+
+from keras.models import load_model
+from keras.preprocessing.image import ImageDataGenerator
+from sklearn.metrics import confusion_matrix, accuracy_score
+
+from myutil import probas_to_classes
+
+
+# Loading and compiling presaved trained CNN
+model = load_model('drawing_classification.h5')
+
+label = {0: "Circle", 1: "Square", 2: "Triangle"}
+
+
+def predict_one(file_name):
+
+    img = cv2.imread(file_name)
+    img = cv2.resize(img, (28, 28))
+    img = np.reshape(img, [1, 28, 28, 3])
+    classes = model.predict_classes(img)[0]
+    category = label[classes]
+    print("\nAnd {1} is the {0}".format(category, file_name))
+    # return category
+
+
+def predict_dataset(input_dir):
+    test_datagen = ImageDataGenerator(rescale=1. / 255)
+    test_generator = test_datagen.flow_from_directory("shapes/test",
+                                                      target_size=(28, 28),
+                                                      color_mode="rgb",
+                                                      shuffle=False,
+                                                      class_mode='categorical',
+                                                      batch_size=1)
+    filenames = test_generator.filenames
+    nb_samples = len(filenames)
+    predict = model.predict_generator(test_generator, steps=nb_samples)
+    return predict, test_generator
+
+
+def main():
+    # Instantiate the parser
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        '--testdata', help='Classify images on test data', action='store_true')
+    parser.add_argument(
+        '--validationdata', help='Classify images on test data', action='store_true')
+
+    parser.add_argument('--image', help='Input your image file name')
+    args = parser.parse_args()
+
+    on_dataset = False
+    if args.testdata:
+        print("Classify images on test dataset")
+        on_dataset = True
+        input_dir = "shapes/test"
+    if args.validationdata:
+        print("Classify images on validation dataset")
+        on_dataset = True
+        input_dir = "shapes/validation"
+
+    if on_dataset:
+        predict, test_generator = predict_dataset(input_dir)
+        y_pred = probas_to_classes(predict)
+        y_true = test_generator.classes
+        X_images = test_generator.filenames
+        cm = confusion_matrix(y_true, y_pred)
+        ac = accuracy_score(y_true, y_pred)
+        for ele in list(zip(X_images, y_true, y_pred)):
+            print(ele)
+    else:
+        file_name = args.image
+        predict_one(file_name)
+
+if __name__ == '__main__':
+    main()