Update app.py

app.py

@@ -180,9 +180,58 @@ def do_prediction(model_name, img):
             label_p_pred = model.predict(img_in, verbose=0)
             num_col = np.argmax(label_p_pred[0]) + 1
             return "Found {} columns".format(num_col), None
+
+        case "SBB/eynollah-page-extraction":
+
+            img_height_model = model.layers[len(model.layers) - 1].output_shape[1]
+            img_width_model = model.layers[len(model.layers) - 1].output_shape[2]
+    
+            img_h_page = img.shape[0]
+            img_w_page = img.shape[1]
+            img = img / float(255.0)
+            img = resize_image(img, img_height_model, img_width_model)
+
+            label_p_pred = model.predict(img.reshape(1, img.shape[0], img.shape[1], img.shape[2]),
+                                         verbose=0)
+
+            seg = np.argmax(label_p_pred, axis=3)[0]
+            seg_color = np.repeat(seg[:, :, np.newaxis], 3, axis=2)
+            prediction_true = resize_image(seg_color, img_h_page, img_w_page)
+            prediction_true = prediction_true.astype(np.uint8)
+            
+            imgray = cv2.cvtColor(prediction_true, cv2.COLOR_BGR2GRAY)
+            _, thresh = cv2.threshold(imgray, 0, 255, 0)
+            #thresh = cv2.dilate(thresh, KERNEL, iterations=3)
+            contours, _ = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
+            
+            if len(contours)>0:
+                cnt_size = np.array([cv2.contourArea(contours[j]) for j in range(len(contours))])
+                cnt = contours[np.argmax(cnt_size)]
+                x, y, w, h = cv2.boundingRect(cnt)
+                if x <= 30:
+                    w += x
+                    x = 0
+                if (img.shape[1] - (x + w)) <= 30:
+                    w = w + (img.shape[1] - (x + w))
+                if y <= 30:
+                    h = h + y
+                    y = 0
+                if (img.shape[0] - (y + h)) <= 30:
+                    h = h + (img.shape[0] - (y + h))
+
+                box = [x, y, w, h]
+
+                img_border = np.zeros((prediction_true.shape[0],prediction_true.shape[1]))
+                img_border[y:y+h, x:x+w] = 1
+            else:
+                img_border = np.zeros((prediction_true.shape[0],prediction_true.shape[1]))
+                img_border[:, :] = 1
+                
+            return "No numerical output", visualize_model_output(img_border,img, model_name)
+                
                             
         # bitmap output
-        case "SBB/eynollah-binarization" | "SBB/eynollah-page-extraction" | "SBB/eynollah-textline" | "SBB/eynollah-textline_light" | "SBB/eynollah-enhancement" | "SBB/eynollah-tables" | "SBB/eynollah-main-regions" | "SBB/eynollah-main-regions-aug-rotation" | "SBB/eynollah-main-regions-aug-scaling" | "SBB/eynollah-main-regions-ensembled" | "SBB/eynollah-full-regions-1column" | "SBB/eynollah-full-regions-3pluscolumn": 
+        case "SBB/eynollah-binarization" | "SBB/eynollah-textline" | "SBB/eynollah-textline_light" | "SBB/eynollah-enhancement" | "SBB/eynollah-tables" | "SBB/eynollah-main-regions" | "SBB/eynollah-main-regions-aug-rotation" | "SBB/eynollah-main-regions-aug-scaling" | "SBB/eynollah-main-regions-ensembled" | "SBB/eynollah-full-regions-1column" | "SBB/eynollah-full-regions-3pluscolumn": 
             
             img_height_model=model.layers[len(model.layers)-1].output_shape[1]
             img_width_model=model.layers[len(model.layers)-1].output_shape[2]
@@ -304,20 +353,6 @@ def do_prediction(model_name, img):
 
             prediction_true = prediction_true.astype(np.uint8)            
 
-            '''
-            img = img / float(255.0)
-            image = resize_image(image, 224,448)
-            prediction = model.predict(image.reshape(1,224,448,image.shape[2]))
-            prediction = tf.squeeze(tf.round(prediction))
-
-            prediction = np.argmax(prediction,axis=2)
-
-            prediction = np.repeat(prediction[:, :, np.newaxis]*255, 3, axis=2)
-            print(prediction.shape)
-            
-            '''
-            #prediction_true = prediction_true * -1
-            #prediction_true = prediction_true + 1
             return "No numerical output", visualize_model_output(prediction_true,img_org, model_name)
         
         # catch-all (we should not reach this)