Practice num 2

practice/1_OpenCV/cat_passport.py

@@ -1,26 +1,38 @@
 import argparse
 import cv2
+import sys
 
 
 def make_cat_passport_image(input_image_path, haar_model_path):
 
     # Read image
-
+    img = cv2.imread(input_image_path)
     # Convert image to grayscale
-
+    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
     # Normalize image intensity
-
+    gray = cv2.equalizeHist(gray)
     # Resize image
 
     # Detect cat faces using Haar Cascade
+    detector = cv2.CascadeClassifier(haar_model_path)
+    rects = detector.detectMultiScale(img, scaleFactor=1.1, minNeighbors=5,
+                                      minSize=(75, 75))
 
     # Draw bounding box
+    for (i, (x, y, w, h)) in enumerate(rects):
+        cv2.rectangle(img, (x, y), (x + w, y + h), (0, 0, 255), 2)
+        cv2.putText(img, f"Cat #{i + 1}", (x, y - 10),
+                    cv2.FONT_HERSHEY_SIMPLEX, 0.55, (0, 0, 255), 2)
 
     # Display result image
-
+    cv2.imshow("window_name", img)
+    cv2.waitKey(0)
+    cv2.destroyAllWindows()
     # Crop image
-
+    x, y, w, h = rects[0]
+    image = img[y:y + h, x:x + w]
     # Save result image to file
+    cv2.imwrite('out.jpg', img)
 
     return
 

practice/2_Classification/doge_classifier.py

@@ -20,32 +20,44 @@ class InferenceEngineClassifier:
     def __init__(self, model_path, device='CPU', classes_path=None):
 
         # Add code for Inference Engine initialization
+        self.core = Core()
 
         # Add code for model loading
+        self.model = self.core.read_model(model=model_path)
 
         # Add code for classes names loading
+        self.exec_model = self.core.compile_model(model=self.model, 
+        device_name=device)
+        if classes_path:     
+            self.classes = [line.rstrip('\n') for line in open(classes_path, encoding='UTF-8')]
 
         return
 
     def get_top(self, prob, topN=1):
-        result = []
-
+        result = prob
+        result = np.squeeze(result)
+        result = np.argsort(result)[-topN:][::-1]
         # Add code for getting top predictions
 
         return result
 
     def _prepare_image(self, image, h, w):
 
         # Add code for image preprocessing
-
+        image = cv2.resize(image, (w, h))
+        image = image.transpose((2, 0, 1))
+        image = np.expand_dims(image, axis = 0)
         return image
 
     def classify(self, image):
-        probabilities = None
-
+        result = None
+        input_layer = self.exec_model.input(0) 
+        output_layer = self.exec_model.output(0)
+        n, c, h, w = input_layer.shape
+        image = self._prepare_image(image, h, w)
         # Add code for image classification using Inference Engine
-
-        return probabilities
+        result = self.exec_model([image])[output_layer]
+        return result
 
 
 def build_argparser():
@@ -71,15 +83,17 @@ def main():
     log.info("Start IE classification sample")
 
     # Create InferenceEngineClassifier object
-
+    test = InferenceEngineClassifier(model_path=args.model, classes_path=args.classes)
     # Read image
-
+    image = cv2.imread(args.input)    
     # Classify image
-
+    classfer = test.classify(image)
     # Get top 5 predictions
-
+    predictions = test.get_top(classfer, 5)
+    classfer = np.squeeze(classfer)
     # print result
-
+    predictions = [str(test.classes[predictions[i]]) + ', ' + str(classfer[predictions[i]]) for i in range(5)]
+    log.info("Predictions: " + str(predictions))
     return