Uploaded models and simple test scripts

jannemus · Jan 10, 2019 · 1c4619e · 1c4619e
1 parent 7fe8314
commit 1c4619e
Show file tree

Hide file tree

Showing 3 changed files with 256 additions and 0 deletions.
diff --git a/models.py b/models.py
@@ -0,0 +1,151 @@
+from keras.models import Model
+from keras.layers import Input, concatenate, Conv2D, Conv2DTranspose, MaxPooling2D, ZeroPadding2D
+from keras import backend as keras
+
+class modelsClass(object):
+
+    def __init__(self, img_rows = 272, img_cols = 480):
+
+        self.img_rows = img_rows
+        self.img_cols = img_cols
+
+    def addPadding(self, layer, level): #height, width, level):
+
+        w1, h1 = self.img_cols, self.img_rows
+        w2, h2 = int(w1/2), int(h1/2)
+        w3, h3 = int(w2/2), int(h2/2)
+        w4, h4 = int(w3/2), int(h3/2)
+        h = [h1,h2,h3,h4]
+        w = [w1,w2,w3,w4]
+
+        # Target width and height
+        tw = w[level-1]
+        th = h[level-1]
+
+        # Source width and height
+        lsize = keras.int_shape(layer)
+        sh = lsize[1]
+        sw = lsize[2]
+
+        pw = (0, tw - sw)
+        ph = (0, th - sh)
+
+        layer = ZeroPadding2D(padding=(ph,pw),data_format="channels_last")(layer)
+
+        return layer
+
+    def getDeepGyro(self):
+
+        input_blurred = Input((self.img_rows, self.img_cols,3))
+        input_blurx = Input((self.img_rows, self.img_cols,1))
+        input_blury = Input((self.img_rows, self.img_cols,1))
+
+        inputs = concatenate([input_blurred,input_blurx,input_blury])
+
+        conv1 = Conv2D(64, (3, 3), activation='relu', padding='same')(inputs)
+        conv1 = Conv2D(64, (3, 3), activation='relu', padding='same')(conv1)
+        pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)
+
+        conv2 = Conv2D(128, (3, 3), activation='relu', padding='same')(pool1)
+        conv2 = Conv2D(128, (3, 3), activation='relu', padding='same')(conv2)
+        pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)
+
+        conv3 = Conv2D(256, (3, 3), activation='relu', padding='same')(pool2)
+        conv3 = Conv2D(256, (3, 3), activation='relu', padding='same')(conv3)
+        pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)
+
+        conv4 = Conv2D(512, (3, 3), activation='relu', padding='same')(pool3)
+        conv4 = Conv2D(512, (3, 3), activation='relu', padding='same')(conv4)
+        pool4 = MaxPooling2D(pool_size=(2, 2))(conv4)
+
+        conv5 = Conv2D(1024, (3, 3), activation='relu', padding='same')(pool4)
+        conv5 = Conv2D(1024, (3, 3), activation='relu', padding='same')(conv5)
+
+        up6 = Conv2DTranspose(512, (2, 2), strides=(2, 2), padding='same')(conv5)
+        up6 = self.addPadding(up6,level=4)
+        up6 = concatenate([up6,conv4], axis=3)
+        conv6 = Conv2D(512, (3, 3), activation='relu', padding='same')(up6)
+        conv6 = Conv2D(512, (3, 3), activation='relu', padding='same')(conv6)
+
+        up7 = Conv2DTranspose(256, (2, 2), strides=(2, 2), padding='same')(conv6)
+        up7 = self.addPadding(up7,level=3)
+        up7 = concatenate([up7,conv3], axis=3)
+        conv7 = Conv2D(256, (3, 3), activation='relu', padding='same')(up7)
+        conv7 = Conv2D(256, (3, 3), activation='relu', padding='same')(conv7)
+
+        up8 = Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same')(conv7)
+        up8 = self.addPadding(up8,level=2)
+        up8 = concatenate([up8,conv2], axis=3)
+        conv8 = Conv2D(128, (3, 3), activation='relu', padding='same')(up8)
+        conv8 = Conv2D(128, (3, 3), activation='relu', padding='same')(conv8)
+
+        up9 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same')(conv8)
+        up9 = self.addPadding(up9,level=1)
+        up9 = concatenate([up9,conv1], axis=3)
+        conv9 = Conv2D(64, (3, 3), activation='relu', padding='same')(up9)
+        conv9 = Conv2D(64, (3, 3), activation='relu', padding='same')(conv9)
+
+        conv10 = Conv2D(3, (1, 1), activation='linear')(conv9)
+
+        model = Model(inputs=[input_blurred,input_blurx,input_blury], outputs=conv10)
+
+        #adam = optimizers.Adam(lr=0.00005, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)
+        #model.compile(optimizer = adam, loss = 'mean_squared_error')
+
+        return model
+
+    def getDeepBlind(self):
+
+        input_blurred = Input((self.img_rows, self.img_cols,3))
+
+        conv1 = Conv2D(64, (3, 3), activation='relu', padding='same')(input_blurred)
+        conv1 = Conv2D(64, (3, 3), activation='relu', padding='same')(conv1)
+        pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)
+
+        conv2 = Conv2D(128, (3, 3), activation='relu', padding='same')(pool1)
+        conv2 = Conv2D(128, (3, 3), activation='relu', padding='same')(conv2)
+        pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)
+
+        conv3 = Conv2D(256, (3, 3), activation='relu', padding='same')(pool2)
+        conv3 = Conv2D(256, (3, 3), activation='relu', padding='same')(conv3)
+        pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)
+
+        conv4 = Conv2D(512, (3, 3), activation='relu', padding='same')(pool3)
+        conv4 = Conv2D(512, (3, 3), activation='relu', padding='same')(conv4)
+        pool4 = MaxPooling2D(pool_size=(2, 2))(conv4)
+
+        conv5 = Conv2D(1024, (3, 3), activation='relu', padding='same')(pool4)
+        conv5 = Conv2D(1024, (3, 3), activation='relu', padding='same')(conv5)
+
+        up6 = Conv2DTranspose(512, (2, 2), strides=(2, 2), padding='same')(conv5)
+        up6 = self.addPadding(up6,level=4)
+        up6 = concatenate([up6,conv4], axis=3)
+        conv6 = Conv2D(512, (3, 3), activation='relu', padding='same')(up6)
+        conv6 = Conv2D(512, (3, 3), activation='relu', padding='same')(conv6)
+
+        up7 = Conv2DTranspose(256, (2, 2), strides=(2, 2), padding='same')(conv6)
+        up7 = self.addPadding(up7,level=3)
+        up7 = concatenate([up7,conv3], axis=3)
+        conv7 = Conv2D(256, (3, 3), activation='relu', padding='same')(up7)
+        conv7 = Conv2D(256, (3, 3), activation='relu', padding='same')(conv7)
+
+        up8 = Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same')(conv7)
+        up8 = self.addPadding(up8,level=2)
+        up8 = concatenate([up8,conv2], axis=3)
+        conv8 = Conv2D(128, (3, 3), activation='relu', padding='same')(up8)
+        conv8 = Conv2D(128, (3, 3), activation='relu', padding='same')(conv8)
+
+        up9 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same')(conv8)
+        up9 = self.addPadding(up9,level=1)
+        up9 = concatenate([up9,conv1], axis=3)
+        conv9 = Conv2D(64, (3, 3), activation='relu', padding='same')(up9)
+        conv9 = Conv2D(64, (3, 3), activation='relu', padding='same')(conv9)
+
+        conv10 = Conv2D(3, (1, 1), activation='linear')(conv9)
+
+        model = Model(inputs=input_blurred, outputs=conv10)
+
+        #adam = optimizers.Adam(lr=0.0000125, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)
+        #model.compile(optimizer = adam, loss = 'mean_squared_error')
+
+        return model
diff --git a/test_deepblind.py b/test_deepblind.py
@@ -0,0 +1,44 @@
+import os, os.path, errno
+from PIL import Image
+import numpy as np
+from keras.models import Model
+from keras.preprocessing.image import array_to_img
+from models import modelsClass
+
+# Put your data to the input folder
+inpath = "input"
+outpath = "output/DeepBlind"
+
+try:
+    os.makedirs(outpath)
+except OSError as e:
+    if e.errno != errno.EEXIST:
+        raise
+
+blurred_path = inpath + "/blurred/blurred/"
+blurred_names = os.listdir(blurred_path)
+
+num_images = len(blurred_names)
+print("Found %d inputs" %(num_images))
+
+for i in range(0, num_images):
+
+    print("Deblurring %s with DeepBlind" %(blurred_names[i]))
+
+    path_blurred = blurred_path + blurred_names[i]       
+    blurred_img = Image.open(path_blurred)
+    blurred_np = (1./255)*np.array(blurred_img)
+
+    width, height = blurred_img.size
+    models = modelsClass(height,width)
+    model = models.getDeepBlind()
+    model.load_weights("checkpoints/DeepBlind.hdf5")
+
+    x = np.reshape(blurred_np,[1,height,width,3])
+    prediction = model.predict(x, batch_size=1,verbose=0,steps=None)
+    prediction = prediction[0,:,:,:]
+
+    deblurred_img = array_to_img(prediction)
+    deblurred_img.save(outpath+"/%s"%(blurred_names[i]))
+
+print("DONE!")
diff --git a/test_deepgyro.py b/test_deepgyro.py
@@ -0,0 +1,61 @@
+import os, os.path, errno
+from PIL import Image
+import numpy as np
+from keras.models import Model
+from keras.preprocessing.image import array_to_img
+from models import modelsClass
+
+# Put your data to the input folder
+inpath = "input"
+outpath = "output/DeepGyro"
+
+try:
+    os.makedirs(outpath)
+except OSError as e:
+    if e.errno != errno.EEXIST:
+        raise
+
+blurred_path = inpath + "/blurred/blurred/"
+blurx_path = inpath + "/blurx/blurx/"
+blury_path = inpath + "/blury/blury/"
+
+blurred_names = os.listdir(blurred_path)
+blurx_names = os.listdir(blurx_path)
+blury_names = os.listdir(blury_path)
+
+num_images = len(blurred_names)
+print("Found %d inputs" %(num_images))
+
+for i in range(0, num_images):
+
+    print("Deblurring %s with DeepGyro" %(blurred_names[i]))
+
+    path_blurred = blurred_path + blurred_names[i]
+    path_blurx = blurx_path + blurx_names[i]
+    path_blury = blury_path + blury_names[i]
+
+    blurred_img = Image.open(path_blurred)
+    blurx_img = Image.open(path_blurx)
+    blury_img = Image.open(path_blury)
+
+    blurred_np = (1./255)*np.array(blurred_img)
+    blurx_np = (1./255)*np.array(blurx_img)
+    blury_np = (1./255)*np.array(blury_img)
+
+    width, height = blurred_img.size
+    models = modelsClass(height,width)
+    model = models.getDeepGyro()
+    model.load_weights("checkpoints/DeepGyro.hdf5")
+
+    b = np.reshape(blurred_np,[1,height,width,3])
+    bx = np.reshape(blurx_np,[1,height,width,1])
+    by = np.reshape(blury_np,[1,height,width,1]) 
+    x = [b,bx,by]
+
+    prediction = model.predict(x, batch_size=1,verbose=0,steps=None)
+    prediction = prediction[0,:,:,:]
+
+    deblurred_img = array_to_img(prediction)
+    deblurred_img.save(outpath+"/%s"%(blurred_names[i]))
+
+print("DONE!")