Panel disconected

Result_ssd7_panel/Panel_Detector.ipynb

@@ -1695,7 +1695,7 @@
     "        current_axis.add_patch(plt.Rectangle((xmin, ymin), xmax-xmin, ymax-ymin, color=color, fill=False, linewidth=2))\n",
     "        current_axis.text(xmin, ymin, label, size='x-large', color='white', bbox={'facecolor':color, 'alpha':1.0})\n",
     "        \n",
-    "        I = orig_images[0]\n",
+    "        I = orig_images[0].copy()\n",
     "        score = np.sum(I[np.int(ymin):np.int(ymax), np.int(xmin):np.int(xmax)]) / ((ymax-ymin)*(xmax-xmin))\n",
     "        \n",
     "        boxes_area_score[str(num)] = {'xmin': xmin, 'xmax': xmax, 'ymin': ymin, 'ymax': ymax, 'score' : score}\n",
@@ -2037,7 +2037,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.6.5"
   }
  },
  "nbformat": 4,

Result_yolo3_panel/.gitignore

@@ -1,5 +1,9 @@
 result_otros/
 log_experimento_panel_gpu/
+Result_Prueba/
+Prueba/
+Otros_2/
+Otros/
 
 *.jpg
 *.h5

config_full_yolo_panel_infer.json

@@ -0,0 +1,49 @@
+{
+    "model" : {
+        "min_input_size":       400,
+        "max_input_size":       400,
+        "anchors":              [5,7, 10,14, 15, 15, 26,32, 45,119, 54,18, 94,59, 109,183, 200,21],
+        "labels":               ["panel"],
+	"backend": 		"keras-yolo3-master/full_yolo_backend.h5"
+    },
+
+    "train": {
+        "train_image_folder":   "Train&Test_A/Train/images/",
+        "train_annot_folder":   "Train&Test_A/Train/anns/",
+	"cache_name":           "Result_yolo3_panel/train_panel.pkl",
+
+        "train_times":          1,
+
+        "batch_size":           2,
+        "learning_rate":        1e-3,
+        "nb_epochs":            500,
+        "warmup_epochs":        15,
+        "ignore_thresh":        0.5,
+        "gpus":                 "0,1",
+
+	"grid_scales":          [1,1,1],
+        "obj_scale":            5,
+        "noobj_scale":          1,
+        "xywh_scale":           1,
+        "class_scale":          1,
+
+	"tensorboard_dir":      "Result_yolo3_panel/log_experimento_panel_gpu",
+	"saved_weights_name":   "Result_yolo3_panel/yolo3_full_panel.h5",
+        "debug":                true
+    },
+
+    "valid": {
+        "valid_image_folder":   "Train&Test_A/Test/images/",
+        "valid_annot_folder":   "Train&Test_A/Test/anns/",
+        "cache_name":           "Result_yolo3_panel/val_panel.pkl",
+
+        "valid_times":          1
+    },
+   "test": {
+        "test_image_folder":   "Train&Test_A/Test/images/",
+        "test_annot_folder":   "Train&Test_A/Test/anns/",
+        "cache_name":          "Result_yolo3_panel/test_panel.pkl",
+
+        "test_times":          1
+    }
+}

panel_disconnect.py

@@ -0,0 +1,41 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Tue Mar 17 13:55:42 2020
+
+@author: dlsaavedra
+"""
+import numpy as np
+
+def disconnect(image, boxes, obj_thresh = 0.5, area_min = 400, merge = 0):
+    
+    new_boxes = []
+    for num, box in enumerate(boxes):
+       
+        xmin = box.xmin + merge
+        xmax = box.xmax - merge
+        ymin = box.ymin + merge
+        ymax = box.ymax - merge
+        
+        if xmin > 0 and ymin > 0 and xmax < image.shape[1] and ymax < image.shape[0] and box.classes[0] > obj_thresh:
+            
+            area = (ymax - ymin)*(xmax - xmin)
+            z_score = np.sum(image[np.int(ymin):np.int(ymax), np.int(xmin):np.int(xmax)]) / area
+            
+            if area > area_min:
+                
+                box.score = z_score
+                new_boxes.append(box)
+                #boxes_area_score[str(num)] = {'xmin': xmin, 'xmax': xmax, 'ymin': ymin, 'ymax': ymax, 'score' : score, 'area' : area}
+       
+    mean_score = np.mean([box.score for box in new_boxes])
+    sd_score  = np.std([box.score for box in new_boxes])
+    
+    new_boxes = [box for box in new_boxes if (box.score - mean_score)/sd_score > 2]
+    
+    for box in new_boxes:
+        
+        z_score = (box.score - mean_score)/sd_score
+        box.classes[0] = min((z_score-2)*0.5+ 0.5, 1)
+        
+    return new_boxes

predict_yolo3.py

@@ -16,7 +16,7 @@ import numpy as np
 
 
 def _main_(args):
-    
+
     config_path  = args.conf
     input_path   = args.input
     output_path  = args.output
@@ -30,7 +30,7 @@ def _main_(args):
     #   Set some parameter
     ###############################
     net_h, net_w = 416, 416 # a multiple of 32, the smaller the faster
-    obj_thresh, nms_thresh = 0.8, 0.3
+    obj_thresh, nms_thresh = 0.5, 0.3
 
     ###############################
     #   Load the model

predict_yolo3_disconnect.py

@@ -15,8 +15,112 @@ from tqdm import tqdm
 import numpy as np
 
 
-def _main_(args):
+def disconnect(image, boxes, obj_thresh = 0.5, area_min = 400, merge = 0):
     
+    new_boxes = []
+    for num, box in enumerate(boxes):
+       
+        xmin = box.xmin + merge
+        xmax = box.xmax - merge
+        ymin = box.ymin + merge
+        ymax = box.ymax - merge
+        
+        if xmin > 0 and ymin > 0 and xmax < image.shape[1] and ymax < image.shape[0] and box.classes[0] > obj_thresh:
+            
+            area = (ymax - ymin)*(xmax - xmin)
+            z_score = np.sum(image[np.int(ymin):np.int(ymax), np.int(xmin):np.int(xmax)]) / area
+            
+            if area > area_min:
+                
+                box.score = z_score
+                new_boxes.append(box)
+                #boxes_area_score[str(num)] = {'xmin': xmin, 'xmax': xmax, 'ymin': ymin, 'ymax': ymax, 'score' : score, 'area' : area}
+       
+    mean_score = np.mean([box.score for box in new_boxes])
+    sd_score  = np.std([box.score for box in new_boxes])
+    
+    new_boxes = [box for box in new_boxes if (box.score - mean_score)/sd_score > 2]
+    
+    for box in new_boxes:
+        
+        z_score = (box.score - mean_score)/sd_score
+        box.classes[0] = min((z_score-2)*0.5+ 0.5, 1)
+        
+    return new_boxes
+    
+
+def disconnect_plot(image, boxes,  obj_thresh = 0.5, area_min = 400, merge = 0):
+        
+    new_boxes = []
+    for num, box in enumerate(boxes):
+       
+        xmin = box.xmin + merge
+        xmax = box.xmax - merge
+        ymin = box.ymin + merge
+        ymax = box.ymax - merge
+        
+        if xmin > 0 and ymin > 0 and xmax < image.shape[1] and ymax < image.shape[0] and box.classes[0] > obj_thresh:
+            
+            area = (ymax - ymin)*(xmax - xmin)
+            z_score = np.sum(image[np.int(ymin):np.int(ymax), np.int(xmin):np.int(xmax)]) / area
+            
+            if area > area_min:
+                
+                box.score = z_score
+                new_boxes.append(box)
+                #boxes_area_score[str(num)] = {'xmin': xmin, 'xmax': xmax, 'ymin': ymin, 'ymax': ymax, 'score' : score, 'area' : area}
+       
+    mean_score = np.mean([box.score for box in new_boxes])
+    sd_score  = np.std([box.score for box in new_boxes])
+    
+     normal_score = ([box.score for box in new_boxes] - mean_score)/sd_score
+#        plt.figure()   
+#         _ = plt.hist(normal_score, bins='auto')  # arguments are passed to np.histogram
+#        plt.title("Histogram with 'auto' bins")
+#        plt.show()
+#        
+#        plt.figure()
+#        mean = np.mean([boxes_area_score[i]['area'] for i in boxes_area_score])
+#        sd  = np.std([boxes_area_score[i]['area'] for i in boxes_area_score])
+#        normal = ([boxes_area_score[i]['area'] for i in boxes_area_score] - mean)/sd
+#        _ = plt.hist(normal, bins='auto')  # arguments are passed to np.histogram
+#        plt.title("Histogram with 'auto' bins")
+#        plt.show()
+    
+    new_boxes = [box for box in new_boxes if (box.score - mean_score)/sd_score > 2]
+    
+    for box in new_boxes:
+        
+        z_score = (box.score - mean_score)/sd_score
+        box.classes[0] = min((z_score-2)*0.5+ 0.5, 1)
+        
+   
+    
+    
+    colors = plt.cm.brg(np.linspace(0, 1, 21)).tolist()
+    plt.figure(figsize=(10,6))
+    plt.imshow(I,cmap = 'gray')
+    current_axis = plt.gca()
+    
+    for box in new_boxes:
+        
+        color = colors[2]
+        
+        #boxes_area_score[key]['score_norm'] = (boxes_area_score[key]['score'] - mean) / sd
+        #z_score = (box.score - mean_score) / sd_score      
+        #z_score = (boxes_area_score[key]['area'] )   
+        
+        ### Escribe el z-score
+        #if z_score > 1:
+        current_axis.text((box.xmin + box.xmax)/2,
+                              (box.ymin+ box.ymax)/2,
+                               '%.2f' % box.classes[0], size='x-large',
+                              color='white', bbox={'facecolor':color, 'alpha':1.0})
+    
+    return new_boxes
+
+def _main_(args):
+
     config_path  = args.conf
     input_path   = args.input
     output_path  = args.output
@@ -122,6 +226,7 @@ def _main_(args):
         times = []
 
         for image_path in image_paths:
+            
             image = cv2.imread(image_path)
             print(image_path)
             start = time.time()
@@ -130,10 +235,9 @@ def _main_(args):
             print('Elapsed time = {}'.format(time.time() - start))
             times.append(time.time() - start)
             # draw bounding boxes on the image using labels
-            for box in boxes:
+            I = image.copy()
-                
+            draw_boxes(I, boxes, config['model']['labels'], obj_thresh)
-            draw_boxes(image, boxes, config['model']['labels'], obj_thresh)
+
-            
             # write the image with bounding boxes to file
             cv2.imwrite(output_path + image_path.split('/')[-1], np.uint8(image))