From 2e91bb570303d062c220d305812bc7f39381bfa1 Mon Sep 17 00:00:00 2001
From: yangfurong09 <71367604+yangfurong09@users.noreply.github.com>
Date: Wed, 16 Sep 2020 17:56:27 +0800
Subject: [PATCH 1/4] Add files via upload

---
 ...05\347\272\247\344\271\235\346\234\210.py" | 425 ++++++++++++++++++
 1 file changed, 425 insertions(+)
 create mode 100644 "1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py"

diff --git "a/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py" "b/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py"
new file mode 100644
index 00000000..c5e78853
--- /dev/null
+++ "b/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py"	
@@ -0,0 +1,425 @@
+#!/usr/bin/env python
+# coding: utf-8
+
+# In[14]:
+
+
+import numpy as np 
+import pandas as pd
+import cv2 
+from PIL import Image
+import os
+import glob
+import xml.etree.ElementTree as ET
+import matplotlib.pyplot as plt
+import matplotlib.patches as patches
+from PIL import Image
+from sklearn.preprocessing import LabelEncoder 
+from sklearn.model_selection import train_test_split
+from seaborn import countplot
+
+data_dir = 'data'
+annon_dir = data_dir + '/annotations'
+imgs_dir = data_dir + '/images'
+SPLIT_RATIO = 0.8 # train and validation split ratio
+snapshot_Folder = 'keras-retinanet/snapshots'
+batch_size = 32
+
+
+# ## 读取数据
+
+# #### Reading data
+
+# In[24]:
+
+
+def read_data():
+    class_names = {}
+    k = 0
+    output = []
+    xml_files = glob.glob("{}/*xml".format(annon_dir))
+    for i, xml_file in enumerate(xml_files):
+        tree = ET.parse(xml_file)
+
+        path = os.path.join(imgs_dir, tree.findtext("./filename"))
+
+        height = int(tree.findtext("./size/height"))
+        width = int(tree.findtext("./size/width"))
+        xmin = int(tree.findtext("./object/bndbox/xmin"))
+        ymin = int(tree.findtext("./object/bndbox/ymin"))
+        xmax = int(tree.findtext("./object/bndbox/xmax"))
+        ymax = int(tree.findtext("./object/bndbox/ymax"))
+
+        basename = os.path.basename(path)
+        basename = os.path.splitext(basename)[0]
+        class_name = basename[:basename.rfind("_")].lower()
+        if class_name not in class_names:
+            class_names[class_name] = k
+            k += 1
+        
+        # making sure bounding boxes are correct
+        if xmin >= xmax or ymin >= ymax or xmax > width or ymax > height or xmin < 0 or ymin < 0:
+            print("Warning: {} contains invalid box. Skipped...".format(path))
+            continue
+            
+        output.append([path, xmin, ymin, xmax, ymax, class_name])
+
+    # preserve percentage of samples for each class ("stratified")
+    output.sort(key=lambda tup : tup[-1])
+    return output
+
+
+# In[25]:
+
+
+data = read_data()
+data[1]
+
+
+# In[27]:
+
+
+# taking a look at data
+
+i = np.random.randint(len(data))
+print(i)
+# A rectangle with uper left at xy = (x, y) with specified width, height and rotation angle.
+im = np.array(Image.open(data[i][0]), dtype=np.uint8)
+# Create figure and axes
+fig,ax = plt.subplots(1)
+# Display the image
+ax.imshow(im)
+
+xleft,yright = data[i][1], data[i][2]
+height = data[i][3]-data[i][1]
+width = data[i][4]-data[i][2]
+print(data[i][-1])
+# print(xleft,yright)
+# print(height,width)
+
+# Create a Rectangle patch
+rect = patches.Rectangle((xleft,yright),
+                         width,
+                         height, linewidth=3, edgecolor='r', facecolor='none')
+# Add the patch to the Axes
+ax.add_patch(rect)
+plt.show()
+
+
+# In[28]:
+
+
+data_df = pd.DataFrame(data,columns=['path','x1','y1','x2','y2','class'])
+data_df.head(3)
+
+
+# In[38]:
+
+
+classes = ['abyssinian', 'american_bulldog', 'american_pit_bull_terrier',
+       'basset_hound', 'beagle', 'bengal', 'birman', 'bombay', 'boxer',
+       'british_shorthair', 'chihuahua', 'egyptian_mau',
+       'english_cocker_spaniel', 'english_setter', 'german_shorthaired',
+       'great_pyrenees', 'havanese', 'japanese_chin', 'keeshond',
+       'leonberger', 'maine_coon', 'miniature_pinscher', 'newfoundland',
+       'persian', 'pomeranian', 'pug', 'ragdoll', 'russian_blue',
+       'saint_bernard', 'samoyed', 'scottish_terrier', 'shiba_inu',
+       'siamese', 'sphynx', 'staffordshire_bull_terrier',
+       'wheaten_terrier', 'yorkshire_terrier']
+
+
+# #### Splitting data
+
+# In[31]:
+
+
+train_df, val_df = train_test_split(data_df, test_size = 0.2, stratify=data_df['class'])
+train_df.head()
+
+
+# In[32]:
+
+
+train_df.to_csv('annotations_train.csv',index=False,header=False)
+val_df.to_csv('annots_valid.csv',index=False,header=False)
+
+
+# **making sure the data is balanced**
+
+# In[33]:
+
+
+countplot(train_df['class'])
+
+
+# In[34]:
+
+
+countplot(val_df['class'])
+
+
+# #### Saving the required csv files for the library to work
+
+# In[39]:
+
+
+classes_df=pd.DataFrame()
+classes_df['class_name']=classes
+classes_df['id']=list(range(0,len(list(classes))))
+classes_df.head()
+
+
+# In[40]:
+
+
+classes_df.to_csv('classes.csv',index=False,header=False)
+
+
+# #### Installing the keras-retinanet library
+
+# https://github.com/fizyr/keras-retinanet#installation
+
+# #### Making the model
+
+# In[ ]:
+
+
+classes = ['abyssinian', 'american_bulldog', 'american_pit_bull_terrier',
+       'basset_hound', 'beagle', 'bengal', 'birman', 'bombay', 'boxer',
+       'british_shorthair', 'chihuahua', 'egyptian_mau',
+       'english_cocker_spaniel', 'english_setter', 'german_shorthaired',
+       'great_pyrenees', 'havanese', 'japanese_chin', 'keeshond',
+       'leonberger', 'maine_coon', 'miniature_pinscher', 'newfoundland',
+       'persian', 'pomeranian', 'pug', 'ragdoll', 'russian_blue',
+       'saint_bernard', 'samoyed', 'scottish_terrier', 'shiba_inu',
+       'siamese', 'sphynx', 'staffordshire_bull_terrier',
+       'wheaten_terrier', 'yorkshire_terrier']
+
+
+# ## 建模 keras
+# 
+
+# In[2]:
+
+
+import matplotlib.pyplot as plt
+from keras_retinanet import keras_retinanet
+from keras_retinanet.keras_retinanet import models
+from keras import optimizers
+from keras_retinanet.keras_retinanet.models import load_model
+from keras_retinanet.keras_retinanet.utils.model import freeze as freeze_model
+from keras_retinanet.keras_retinanet.preprocessing import csv_generator
+
+
+# In[3]:
+
+
+# making themodel and loading the weights of the model
+model = models.backbone('resnet50').retinanet(num_classes=len(classes),modifier=freeze_model)
+model.load_weights('keras_retinanet/snapshots/resnet50_coco_best_v2.1.0.h5', by_name=True,skip_mismatch=True)
+model.compile(
+    loss={
+        'regression'    : keras_retinanet.losses.smooth_l1(),
+        'classification': keras_retinanet.losses.focal()
+    },
+    optimizer=optimizers.adam(lr=1e-5, clipnorm=0.001)
+)
+
+
+# In[4]:
+
+
+model.summary()
+
+
+# In[2]:
+
+
+# 训练
+
+batch_size = 4
+data_generator = csv_generator.CSVGenerator(csv_data_file='annotations_train.csv',csv_class_file='classes.csv',
+                              batch_size=batch_size)
+data_generator_val = csv_generator.CSVGenerator(csv_data_file='annots_valid.csv',csv_class_file='classes.csv', 
+                              batch_size=batch_size)
+
+snapshot_folder = 'weights'
+history={'losses_val':[], 'losses':[]}
+ep = 1
+
+
+# In[6]:
+
+
+# model.evaluate_generator(data_generator_val, workers=8, use_multiprocessing=False, verbose=1)
+
+
+# In[13]:
+
+
+def evaluate():
+    l,r_l,c_l = model.evaluate_generator(data_generator_val, workers=8, use_multiprocessing=False)
+#     print('Total loss: ',l,' Regression loss: ', r_l,' classification loss: ',c_l)
+    return l,r_l,c_l
+
+
+# * freeze-backbone: freeze the backbone layers, particularly useful when we use a small dataset, to avoid overfitting
+# * random-transform: randomly transform the dataset to get data augmentation
+# * weights: initialize the model with a pretrained model (your own model or one released by Fizyr)
+# * batch-size: training batch size, higher value gives smoother learning curve
+# * steps: number of steps for epochs
+# * epochs: number of epochs to train
+# * csv: annotations files generated by the script above
+
+# In[8]:
+
+
+import pandas as pd
+training_samples = len(pd.read_csv('annotations_train.csv'))
+training_samples
+
+
+# In[9]:
+
+
+import numpy as np
+
+
+# In[38]:
+
+
+epochs = 200
+steps = training_samples//batch_size
+
+for i in range(ep,epochs):
+    
+    print('Epoch :',i,'\n')
+    # fit for one epoch
+    h = model.fit_generator(data_generator, epochs=1, steps_per_epoch=steps, verbose=1)
+    ep = i + 1
+    history['losses'].append(h.history)
+    model.save(snapshot_folder+'/model_' + str(i) + '.h5')
+    # save model every 10 epochs
+    if i % 3 == 0:
+        evaluate()
+        history['losses_val'].append((evaluate(),i))
+        plot_history()
+        
+    print('\n','='*80)
+
+
+# In[37]:
+
+
+model.optimizer.learning_rate = model.optimizer.learning_rate/10 
+
+
+# In[33]:
+
+
+def plot_history(loss='losses_val'):
+    total_val_loss = [i[0] for i in history[loss]]
+    plt.figure(figsize=(4,3))
+    plt.plot([i*3 for i in range(1,1+len(total_val_loss))],total_val_loss,'ro-')
+    plt.title(loss)
+    plt.show()
+    
+plot_history()
+
+
+# In[108]:
+
+
+get_ipython().run_line_magic('matplotlib', 'inline')
+
+# automatically reload modules when they have changed
+get_ipython().run_line_magic('load_ext', 'autoreload')
+get_ipython().run_line_magic('autoreload', '2')
+
+# import keras_retinanet
+from keras_retinanet.keras_retinanet.utils.image import read_image_bgr, preprocess_image, resize_image
+from keras_retinanet.keras_retinanet.utils.visualization import draw_box, draw_caption
+from keras_retinanet.keras_retinanet.utils.colors import label_color
+
+
+# In[109]:
+
+
+model_inf = models.load_model(snapshot_folder+'/model_' + str(9) + '.h5', backbone_name='resnet50')
+model_inf = models.convert_model(model) # converting to inference mode
+
+
+# In[111]:
+
+
+classes_dic={}
+i=0
+for c in classes:
+    classes_dic[i]=c
+    i+=1
+
+
+# In[112]:
+
+
+def test(loc,thresh=0.5):
+    # load image
+    image = read_image_bgr(loc)
+    # copy to draw on
+    draw = image.copy()
+    draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
+
+    # preprocess image for network
+    image = preprocess_image(image)
+    image, scale = resize_image(image)
+    # process image
+    start = time.time()
+    boxes, scores, labels = model_inf.predict_on_batch(np.expand_dims(image, axis=0))
+    print("processing time: ", time.time() - start)
+
+    # correct for image scale
+    boxes /= scale
+
+    # visualize detections
+    for box, score, label in zip(boxes[0], scores[0], labels[0]):
+        # scores are sorted so we can break
+        if score < thresh:
+            break
+
+        color = label_color(label)
+
+        b = box.astype(int)
+        draw_box(draw, b, color=color)
+
+        caption = "{} {:.3f}".format(classes_dic[label], score)
+        draw_caption(draw, b, caption)
+
+    plt.figure(figsize=(15, 15))
+    plt.axis('off')
+    plt.imshow(draw)
+    plt.show()
+
+
+# In[117]:
+
+
+
+
+# load image
+index = np.random.randint(1,len(val_data))
+print(index)
+test(val_data.loc[index][0],0.7)
+
+
+# In[115]:
+
+
+test('sample/img1.jpg')
+test('sample/img2.jpg',0.2)
+
+
+# In[ ]:
+
+
+
+

From 01cad3d7d3489da64c5b7b47d386ab37b105bf68 Mon Sep 17 00:00:00 2001
From: yangfurong09 <71367604+yangfurong09@users.noreply.github.com>
Date: Wed, 16 Sep 2020 21:23:27 +0800
Subject: [PATCH 2/4] Add files via upload


From 23e86ff233c6d8afb043293e968bc0620066d40a Mon Sep 17 00:00:00 2001
From: yangfurong09 <71367604+yangfurong09@users.noreply.github.com>
Date: Wed, 16 Sep 2020 21:26:41 +0800
Subject: [PATCH 3/4] =?UTF-8?q?Delete=20=E8=B6=85=E7=BA=A7=E4=B9=9D?=
 =?UTF-8?q?=E6=9C=88.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 ...05\347\272\247\344\271\235\346\234\210.py" | 425 ------------------
 1 file changed, 425 deletions(-)
 delete mode 100644 "1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py"

diff --git "a/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py" "b/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py"
deleted file mode 100644
index c5e78853..00000000
--- "a/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\350\266\205\347\272\247\344\271\235\346\234\210.py"	
+++ /dev/null
@@ -1,425 +0,0 @@
-#!/usr/bin/env python
-# coding: utf-8
-
-# In[14]:
-
-
-import numpy as np 
-import pandas as pd
-import cv2 
-from PIL import Image
-import os
-import glob
-import xml.etree.ElementTree as ET
-import matplotlib.pyplot as plt
-import matplotlib.patches as patches
-from PIL import Image
-from sklearn.preprocessing import LabelEncoder 
-from sklearn.model_selection import train_test_split
-from seaborn import countplot
-
-data_dir = 'data'
-annon_dir = data_dir + '/annotations'
-imgs_dir = data_dir + '/images'
-SPLIT_RATIO = 0.8 # train and validation split ratio
-snapshot_Folder = 'keras-retinanet/snapshots'
-batch_size = 32
-
-
-# ## 读取数据
-
-# #### Reading data
-
-# In[24]:
-
-
-def read_data():
-    class_names = {}
-    k = 0
-    output = []
-    xml_files = glob.glob("{}/*xml".format(annon_dir))
-    for i, xml_file in enumerate(xml_files):
-        tree = ET.parse(xml_file)
-
-        path = os.path.join(imgs_dir, tree.findtext("./filename"))
-
-        height = int(tree.findtext("./size/height"))
-        width = int(tree.findtext("./size/width"))
-        xmin = int(tree.findtext("./object/bndbox/xmin"))
-        ymin = int(tree.findtext("./object/bndbox/ymin"))
-        xmax = int(tree.findtext("./object/bndbox/xmax"))
-        ymax = int(tree.findtext("./object/bndbox/ymax"))
-
-        basename = os.path.basename(path)
-        basename = os.path.splitext(basename)[0]
-        class_name = basename[:basename.rfind("_")].lower()
-        if class_name not in class_names:
-            class_names[class_name] = k
-            k += 1
-        
-        # making sure bounding boxes are correct
-        if xmin >= xmax or ymin >= ymax or xmax > width or ymax > height or xmin < 0 or ymin < 0:
-            print("Warning: {} contains invalid box. Skipped...".format(path))
-            continue
-            
-        output.append([path, xmin, ymin, xmax, ymax, class_name])
-
-    # preserve percentage of samples for each class ("stratified")
-    output.sort(key=lambda tup : tup[-1])
-    return output
-
-
-# In[25]:
-
-
-data = read_data()
-data[1]
-
-
-# In[27]:
-
-
-# taking a look at data
-
-i = np.random.randint(len(data))
-print(i)
-# A rectangle with uper left at xy = (x, y) with specified width, height and rotation angle.
-im = np.array(Image.open(data[i][0]), dtype=np.uint8)
-# Create figure and axes
-fig,ax = plt.subplots(1)
-# Display the image
-ax.imshow(im)
-
-xleft,yright = data[i][1], data[i][2]
-height = data[i][3]-data[i][1]
-width = data[i][4]-data[i][2]
-print(data[i][-1])
-# print(xleft,yright)
-# print(height,width)
-
-# Create a Rectangle patch
-rect = patches.Rectangle((xleft,yright),
-                         width,
-                         height, linewidth=3, edgecolor='r', facecolor='none')
-# Add the patch to the Axes
-ax.add_patch(rect)
-plt.show()
-
-
-# In[28]:
-
-
-data_df = pd.DataFrame(data,columns=['path','x1','y1','x2','y2','class'])
-data_df.head(3)
-
-
-# In[38]:
-
-
-classes = ['abyssinian', 'american_bulldog', 'american_pit_bull_terrier',
-       'basset_hound', 'beagle', 'bengal', 'birman', 'bombay', 'boxer',
-       'british_shorthair', 'chihuahua', 'egyptian_mau',
-       'english_cocker_spaniel', 'english_setter', 'german_shorthaired',
-       'great_pyrenees', 'havanese', 'japanese_chin', 'keeshond',
-       'leonberger', 'maine_coon', 'miniature_pinscher', 'newfoundland',
-       'persian', 'pomeranian', 'pug', 'ragdoll', 'russian_blue',
-       'saint_bernard', 'samoyed', 'scottish_terrier', 'shiba_inu',
-       'siamese', 'sphynx', 'staffordshire_bull_terrier',
-       'wheaten_terrier', 'yorkshire_terrier']
-
-
-# #### Splitting data
-
-# In[31]:
-
-
-train_df, val_df = train_test_split(data_df, test_size = 0.2, stratify=data_df['class'])
-train_df.head()
-
-
-# In[32]:
-
-
-train_df.to_csv('annotations_train.csv',index=False,header=False)
-val_df.to_csv('annots_valid.csv',index=False,header=False)
-
-
-# **making sure the data is balanced**
-
-# In[33]:
-
-
-countplot(train_df['class'])
-
-
-# In[34]:
-
-
-countplot(val_df['class'])
-
-
-# #### Saving the required csv files for the library to work
-
-# In[39]:
-
-
-classes_df=pd.DataFrame()
-classes_df['class_name']=classes
-classes_df['id']=list(range(0,len(list(classes))))
-classes_df.head()
-
-
-# In[40]:
-
-
-classes_df.to_csv('classes.csv',index=False,header=False)
-
-
-# #### Installing the keras-retinanet library
-
-# https://github.com/fizyr/keras-retinanet#installation
-
-# #### Making the model
-
-# In[ ]:
-
-
-classes = ['abyssinian', 'american_bulldog', 'american_pit_bull_terrier',
-       'basset_hound', 'beagle', 'bengal', 'birman', 'bombay', 'boxer',
-       'british_shorthair', 'chihuahua', 'egyptian_mau',
-       'english_cocker_spaniel', 'english_setter', 'german_shorthaired',
-       'great_pyrenees', 'havanese', 'japanese_chin', 'keeshond',
-       'leonberger', 'maine_coon', 'miniature_pinscher', 'newfoundland',
-       'persian', 'pomeranian', 'pug', 'ragdoll', 'russian_blue',
-       'saint_bernard', 'samoyed', 'scottish_terrier', 'shiba_inu',
-       'siamese', 'sphynx', 'staffordshire_bull_terrier',
-       'wheaten_terrier', 'yorkshire_terrier']
-
-
-# ## 建模 keras
-# 
-
-# In[2]:
-
-
-import matplotlib.pyplot as plt
-from keras_retinanet import keras_retinanet
-from keras_retinanet.keras_retinanet import models
-from keras import optimizers
-from keras_retinanet.keras_retinanet.models import load_model
-from keras_retinanet.keras_retinanet.utils.model import freeze as freeze_model
-from keras_retinanet.keras_retinanet.preprocessing import csv_generator
-
-
-# In[3]:
-
-
-# making themodel and loading the weights of the model
-model = models.backbone('resnet50').retinanet(num_classes=len(classes),modifier=freeze_model)
-model.load_weights('keras_retinanet/snapshots/resnet50_coco_best_v2.1.0.h5', by_name=True,skip_mismatch=True)
-model.compile(
-    loss={
-        'regression'    : keras_retinanet.losses.smooth_l1(),
-        'classification': keras_retinanet.losses.focal()
-    },
-    optimizer=optimizers.adam(lr=1e-5, clipnorm=0.001)
-)
-
-
-# In[4]:
-
-
-model.summary()
-
-
-# In[2]:
-
-
-# 训练
-
-batch_size = 4
-data_generator = csv_generator.CSVGenerator(csv_data_file='annotations_train.csv',csv_class_file='classes.csv',
-                              batch_size=batch_size)
-data_generator_val = csv_generator.CSVGenerator(csv_data_file='annots_valid.csv',csv_class_file='classes.csv', 
-                              batch_size=batch_size)
-
-snapshot_folder = 'weights'
-history={'losses_val':[], 'losses':[]}
-ep = 1
-
-
-# In[6]:
-
-
-# model.evaluate_generator(data_generator_val, workers=8, use_multiprocessing=False, verbose=1)
-
-
-# In[13]:
-
-
-def evaluate():
-    l,r_l,c_l = model.evaluate_generator(data_generator_val, workers=8, use_multiprocessing=False)
-#     print('Total loss: ',l,' Regression loss: ', r_l,' classification loss: ',c_l)
-    return l,r_l,c_l
-
-
-# * freeze-backbone: freeze the backbone layers, particularly useful when we use a small dataset, to avoid overfitting
-# * random-transform: randomly transform the dataset to get data augmentation
-# * weights: initialize the model with a pretrained model (your own model or one released by Fizyr)
-# * batch-size: training batch size, higher value gives smoother learning curve
-# * steps: number of steps for epochs
-# * epochs: number of epochs to train
-# * csv: annotations files generated by the script above
-
-# In[8]:
-
-
-import pandas as pd
-training_samples = len(pd.read_csv('annotations_train.csv'))
-training_samples
-
-
-# In[9]:
-
-
-import numpy as np
-
-
-# In[38]:
-
-
-epochs = 200
-steps = training_samples//batch_size
-
-for i in range(ep,epochs):
-    
-    print('Epoch :',i,'\n')
-    # fit for one epoch
-    h = model.fit_generator(data_generator, epochs=1, steps_per_epoch=steps, verbose=1)
-    ep = i + 1
-    history['losses'].append(h.history)
-    model.save(snapshot_folder+'/model_' + str(i) + '.h5')
-    # save model every 10 epochs
-    if i % 3 == 0:
-        evaluate()
-        history['losses_val'].append((evaluate(),i))
-        plot_history()
-        
-    print('\n','='*80)
-
-
-# In[37]:
-
-
-model.optimizer.learning_rate = model.optimizer.learning_rate/10 
-
-
-# In[33]:
-
-
-def plot_history(loss='losses_val'):
-    total_val_loss = [i[0] for i in history[loss]]
-    plt.figure(figsize=(4,3))
-    plt.plot([i*3 for i in range(1,1+len(total_val_loss))],total_val_loss,'ro-')
-    plt.title(loss)
-    plt.show()
-    
-plot_history()
-
-
-# In[108]:
-
-
-get_ipython().run_line_magic('matplotlib', 'inline')
-
-# automatically reload modules when they have changed
-get_ipython().run_line_magic('load_ext', 'autoreload')
-get_ipython().run_line_magic('autoreload', '2')
-
-# import keras_retinanet
-from keras_retinanet.keras_retinanet.utils.image import read_image_bgr, preprocess_image, resize_image
-from keras_retinanet.keras_retinanet.utils.visualization import draw_box, draw_caption
-from keras_retinanet.keras_retinanet.utils.colors import label_color
-
-
-# In[109]:
-
-
-model_inf = models.load_model(snapshot_folder+'/model_' + str(9) + '.h5', backbone_name='resnet50')
-model_inf = models.convert_model(model) # converting to inference mode
-
-
-# In[111]:
-
-
-classes_dic={}
-i=0
-for c in classes:
-    classes_dic[i]=c
-    i+=1
-
-
-# In[112]:
-
-
-def test(loc,thresh=0.5):
-    # load image
-    image = read_image_bgr(loc)
-    # copy to draw on
-    draw = image.copy()
-    draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
-
-    # preprocess image for network
-    image = preprocess_image(image)
-    image, scale = resize_image(image)
-    # process image
-    start = time.time()
-    boxes, scores, labels = model_inf.predict_on_batch(np.expand_dims(image, axis=0))
-    print("processing time: ", time.time() - start)
-
-    # correct for image scale
-    boxes /= scale
-
-    # visualize detections
-    for box, score, label in zip(boxes[0], scores[0], labels[0]):
-        # scores are sorted so we can break
-        if score < thresh:
-            break
-
-        color = label_color(label)
-
-        b = box.astype(int)
-        draw_box(draw, b, color=color)
-
-        caption = "{} {:.3f}".format(classes_dic[label], score)
-        draw_caption(draw, b, caption)
-
-    plt.figure(figsize=(15, 15))
-    plt.axis('off')
-    plt.imshow(draw)
-    plt.show()
-
-
-# In[117]:
-
-
-
-
-# load image
-index = np.random.randint(1,len(val_data))
-print(index)
-test(val_data.loc[index][0],0.7)
-
-
-# In[115]:
-
-
-test('sample/img1.jpg')
-test('sample/img2.jpg',0.2)
-
-
-# In[ ]:
-
-
-
-

From 761d483ab760327c2c1b31e696dcfbefcdfb480d Mon Sep 17 00:00:00 2001
From: yangfurong09 <71367604+yangfurong09@users.noreply.github.com>
Date: Wed, 16 Sep 2020 21:28:11 +0800
Subject: [PATCH 4/4] Add files via upload

---
 ...05\347\272\247\344\271\235\346\234\210.py" | 425 ++++++++++++++++++
 1 file changed, 425 insertions(+)
 create mode 100644 "1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\344\271\235\346\234\210\345\245\207\350\277\271/\350\266\205\347\272\247\344\271\235\346\234\210.py"

diff --git "a/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\344\271\235\346\234\210\345\245\207\350\277\271/\350\266\205\347\272\247\344\271\235\346\234\210.py" "b/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\344\271\235\346\234\210\345\245\207\350\277\271/\350\266\205\347\272\247\344\271\235\346\234\210.py"
new file mode 100644
index 00000000..c5e78853
--- /dev/null
+++ "b/1 \351\201\227\345\244\261\345\256\240\347\211\251\347\232\204\346\231\272\350\203\275\345\257\273\346\211\276/\344\271\235\346\234\210\345\245\207\350\277\271/\350\266\205\347\272\247\344\271\235\346\234\210.py"	
@@ -0,0 +1,425 @@
+#!/usr/bin/env python
+# coding: utf-8
+
+# In[14]:
+
+
+import numpy as np 
+import pandas as pd
+import cv2 
+from PIL import Image
+import os
+import glob
+import xml.etree.ElementTree as ET
+import matplotlib.pyplot as plt
+import matplotlib.patches as patches
+from PIL import Image
+from sklearn.preprocessing import LabelEncoder 
+from sklearn.model_selection import train_test_split
+from seaborn import countplot
+
+data_dir = 'data'
+annon_dir = data_dir + '/annotations'
+imgs_dir = data_dir + '/images'
+SPLIT_RATIO = 0.8 # train and validation split ratio
+snapshot_Folder = 'keras-retinanet/snapshots'
+batch_size = 32
+
+
+# ## 读取数据
+
+# #### Reading data
+
+# In[24]:
+
+
+def read_data():
+    class_names = {}
+    k = 0
+    output = []
+    xml_files = glob.glob("{}/*xml".format(annon_dir))
+    for i, xml_file in enumerate(xml_files):
+        tree = ET.parse(xml_file)
+
+        path = os.path.join(imgs_dir, tree.findtext("./filename"))
+
+        height = int(tree.findtext("./size/height"))
+        width = int(tree.findtext("./size/width"))
+        xmin = int(tree.findtext("./object/bndbox/xmin"))
+        ymin = int(tree.findtext("./object/bndbox/ymin"))
+        xmax = int(tree.findtext("./object/bndbox/xmax"))
+        ymax = int(tree.findtext("./object/bndbox/ymax"))
+
+        basename = os.path.basename(path)
+        basename = os.path.splitext(basename)[0]
+        class_name = basename[:basename.rfind("_")].lower()
+        if class_name not in class_names:
+            class_names[class_name] = k
+            k += 1
+        
+        # making sure bounding boxes are correct
+        if xmin >= xmax or ymin >= ymax or xmax > width or ymax > height or xmin < 0 or ymin < 0:
+            print("Warning: {} contains invalid box. Skipped...".format(path))
+            continue
+            
+        output.append([path, xmin, ymin, xmax, ymax, class_name])
+
+    # preserve percentage of samples for each class ("stratified")
+    output.sort(key=lambda tup : tup[-1])
+    return output
+
+
+# In[25]:
+
+
+data = read_data()
+data[1]
+
+
+# In[27]:
+
+
+# taking a look at data
+
+i = np.random.randint(len(data))
+print(i)
+# A rectangle with uper left at xy = (x, y) with specified width, height and rotation angle.
+im = np.array(Image.open(data[i][0]), dtype=np.uint8)
+# Create figure and axes
+fig,ax = plt.subplots(1)
+# Display the image
+ax.imshow(im)
+
+xleft,yright = data[i][1], data[i][2]
+height = data[i][3]-data[i][1]
+width = data[i][4]-data[i][2]
+print(data[i][-1])
+# print(xleft,yright)
+# print(height,width)
+
+# Create a Rectangle patch
+rect = patches.Rectangle((xleft,yright),
+                         width,
+                         height, linewidth=3, edgecolor='r', facecolor='none')
+# Add the patch to the Axes
+ax.add_patch(rect)
+plt.show()
+
+
+# In[28]:
+
+
+data_df = pd.DataFrame(data,columns=['path','x1','y1','x2','y2','class'])
+data_df.head(3)
+
+
+# In[38]:
+
+
+classes = ['abyssinian', 'american_bulldog', 'american_pit_bull_terrier',
+       'basset_hound', 'beagle', 'bengal', 'birman', 'bombay', 'boxer',
+       'british_shorthair', 'chihuahua', 'egyptian_mau',
+       'english_cocker_spaniel', 'english_setter', 'german_shorthaired',
+       'great_pyrenees', 'havanese', 'japanese_chin', 'keeshond',
+       'leonberger', 'maine_coon', 'miniature_pinscher', 'newfoundland',
+       'persian', 'pomeranian', 'pug', 'ragdoll', 'russian_blue',
+       'saint_bernard', 'samoyed', 'scottish_terrier', 'shiba_inu',
+       'siamese', 'sphynx', 'staffordshire_bull_terrier',
+       'wheaten_terrier', 'yorkshire_terrier']
+
+
+# #### Splitting data
+
+# In[31]:
+
+
+train_df, val_df = train_test_split(data_df, test_size = 0.2, stratify=data_df['class'])
+train_df.head()
+
+
+# In[32]:
+
+
+train_df.to_csv('annotations_train.csv',index=False,header=False)
+val_df.to_csv('annots_valid.csv',index=False,header=False)
+
+
+# **making sure the data is balanced**
+
+# In[33]:
+
+
+countplot(train_df['class'])
+
+
+# In[34]:
+
+
+countplot(val_df['class'])
+
+
+# #### Saving the required csv files for the library to work
+
+# In[39]:
+
+
+classes_df=pd.DataFrame()
+classes_df['class_name']=classes
+classes_df['id']=list(range(0,len(list(classes))))
+classes_df.head()
+
+
+# In[40]:
+
+
+classes_df.to_csv('classes.csv',index=False,header=False)
+
+
+# #### Installing the keras-retinanet library
+
+# https://github.com/fizyr/keras-retinanet#installation
+
+# #### Making the model
+
+# In[ ]:
+
+
+classes = ['abyssinian', 'american_bulldog', 'american_pit_bull_terrier',
+       'basset_hound', 'beagle', 'bengal', 'birman', 'bombay', 'boxer',
+       'british_shorthair', 'chihuahua', 'egyptian_mau',
+       'english_cocker_spaniel', 'english_setter', 'german_shorthaired',
+       'great_pyrenees', 'havanese', 'japanese_chin', 'keeshond',
+       'leonberger', 'maine_coon', 'miniature_pinscher', 'newfoundland',
+       'persian', 'pomeranian', 'pug', 'ragdoll', 'russian_blue',
+       'saint_bernard', 'samoyed', 'scottish_terrier', 'shiba_inu',
+       'siamese', 'sphynx', 'staffordshire_bull_terrier',
+       'wheaten_terrier', 'yorkshire_terrier']
+
+
+# ## 建模 keras
+# 
+
+# In[2]:
+
+
+import matplotlib.pyplot as plt
+from keras_retinanet import keras_retinanet
+from keras_retinanet.keras_retinanet import models
+from keras import optimizers
+from keras_retinanet.keras_retinanet.models import load_model
+from keras_retinanet.keras_retinanet.utils.model import freeze as freeze_model
+from keras_retinanet.keras_retinanet.preprocessing import csv_generator
+
+
+# In[3]:
+
+
+# making themodel and loading the weights of the model
+model = models.backbone('resnet50').retinanet(num_classes=len(classes),modifier=freeze_model)
+model.load_weights('keras_retinanet/snapshots/resnet50_coco_best_v2.1.0.h5', by_name=True,skip_mismatch=True)
+model.compile(
+    loss={
+        'regression'    : keras_retinanet.losses.smooth_l1(),
+        'classification': keras_retinanet.losses.focal()
+    },
+    optimizer=optimizers.adam(lr=1e-5, clipnorm=0.001)
+)
+
+
+# In[4]:
+
+
+model.summary()
+
+
+# In[2]:
+
+
+# 训练
+
+batch_size = 4
+data_generator = csv_generator.CSVGenerator(csv_data_file='annotations_train.csv',csv_class_file='classes.csv',
+                              batch_size=batch_size)
+data_generator_val = csv_generator.CSVGenerator(csv_data_file='annots_valid.csv',csv_class_file='classes.csv', 
+                              batch_size=batch_size)
+
+snapshot_folder = 'weights'
+history={'losses_val':[], 'losses':[]}
+ep = 1
+
+
+# In[6]:
+
+
+# model.evaluate_generator(data_generator_val, workers=8, use_multiprocessing=False, verbose=1)
+
+
+# In[13]:
+
+
+def evaluate():
+    l,r_l,c_l = model.evaluate_generator(data_generator_val, workers=8, use_multiprocessing=False)
+#     print('Total loss: ',l,' Regression loss: ', r_l,' classification loss: ',c_l)
+    return l,r_l,c_l
+
+
+# * freeze-backbone: freeze the backbone layers, particularly useful when we use a small dataset, to avoid overfitting
+# * random-transform: randomly transform the dataset to get data augmentation
+# * weights: initialize the model with a pretrained model (your own model or one released by Fizyr)
+# * batch-size: training batch size, higher value gives smoother learning curve
+# * steps: number of steps for epochs
+# * epochs: number of epochs to train
+# * csv: annotations files generated by the script above
+
+# In[8]:
+
+
+import pandas as pd
+training_samples = len(pd.read_csv('annotations_train.csv'))
+training_samples
+
+
+# In[9]:
+
+
+import numpy as np
+
+
+# In[38]:
+
+
+epochs = 200
+steps = training_samples//batch_size
+
+for i in range(ep,epochs):
+    
+    print('Epoch :',i,'\n')
+    # fit for one epoch
+    h = model.fit_generator(data_generator, epochs=1, steps_per_epoch=steps, verbose=1)
+    ep = i + 1
+    history['losses'].append(h.history)
+    model.save(snapshot_folder+'/model_' + str(i) + '.h5')
+    # save model every 10 epochs
+    if i % 3 == 0:
+        evaluate()
+        history['losses_val'].append((evaluate(),i))
+        plot_history()
+        
+    print('\n','='*80)
+
+
+# In[37]:
+
+
+model.optimizer.learning_rate = model.optimizer.learning_rate/10 
+
+
+# In[33]:
+
+
+def plot_history(loss='losses_val'):
+    total_val_loss = [i[0] for i in history[loss]]
+    plt.figure(figsize=(4,3))
+    plt.plot([i*3 for i in range(1,1+len(total_val_loss))],total_val_loss,'ro-')
+    plt.title(loss)
+    plt.show()
+    
+plot_history()
+
+
+# In[108]:
+
+
+get_ipython().run_line_magic('matplotlib', 'inline')
+
+# automatically reload modules when they have changed
+get_ipython().run_line_magic('load_ext', 'autoreload')
+get_ipython().run_line_magic('autoreload', '2')
+
+# import keras_retinanet
+from keras_retinanet.keras_retinanet.utils.image import read_image_bgr, preprocess_image, resize_image
+from keras_retinanet.keras_retinanet.utils.visualization import draw_box, draw_caption
+from keras_retinanet.keras_retinanet.utils.colors import label_color
+
+
+# In[109]:
+
+
+model_inf = models.load_model(snapshot_folder+'/model_' + str(9) + '.h5', backbone_name='resnet50')
+model_inf = models.convert_model(model) # converting to inference mode
+
+
+# In[111]:
+
+
+classes_dic={}
+i=0
+for c in classes:
+    classes_dic[i]=c
+    i+=1
+
+
+# In[112]:
+
+
+def test(loc,thresh=0.5):
+    # load image
+    image = read_image_bgr(loc)
+    # copy to draw on
+    draw = image.copy()
+    draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
+
+    # preprocess image for network
+    image = preprocess_image(image)
+    image, scale = resize_image(image)
+    # process image
+    start = time.time()
+    boxes, scores, labels = model_inf.predict_on_batch(np.expand_dims(image, axis=0))
+    print("processing time: ", time.time() - start)
+
+    # correct for image scale
+    boxes /= scale
+
+    # visualize detections
+    for box, score, label in zip(boxes[0], scores[0], labels[0]):
+        # scores are sorted so we can break
+        if score < thresh:
+            break
+
+        color = label_color(label)
+
+        b = box.astype(int)
+        draw_box(draw, b, color=color)
+
+        caption = "{} {:.3f}".format(classes_dic[label], score)
+        draw_caption(draw, b, caption)
+
+    plt.figure(figsize=(15, 15))
+    plt.axis('off')
+    plt.imshow(draw)
+    plt.show()
+
+
+# In[117]:
+
+
+
+
+# load image
+index = np.random.randint(1,len(val_data))
+print(index)
+test(val_data.loc[index][0],0.7)
+
+
+# In[115]:
+
+
+test('sample/img1.jpg')
+test('sample/img2.jpg',0.2)
+
+
+# In[ ]:
+
+
+
+