def image_classification(paths):
classification_dict = {}
model = MobileNet(input_shape=None, alpha=1.0, depth_multiplier=1, include_top=True, weights='imagenet', input_tensor=None, pooling=None, classes=1000)
for path in paths:
image = mnetv2_loadimage(path)
preds = model.predict(image)
prediction = decode_predictions(preds, top=1)[0][0]
img_class = prediction[1]
img_confidence = prediction[2]
if img_class in classification_dict:
classification_dict[img_class].append((path, img_confidence))
else:
classification_dict[img_class] = [(path, img_confidence)]
return classification_dict
class MobileNetWrapper(object):
def __init__(self):
logger.info('Loading MobileNet')
self.model = MobileNet(weights='imagenet')
def predict(self, img):
""" # Arguments
img: a numpy array
# Returns
A dict containing predictions
"""
img = Image.fromarray(img)
img = img.resize((224, 224))
x = keras_image.img_to_array(img)[:, :, :3]
x = np.expand_dims(x, axis=0)
x = preprocess_input_mobilenet(x)
features = self.model.predict(x)
predictions = decode_predictions_mobilenet(features)[0]
clean_predictions = [{'score': str(k), 'class': j} for (i, j, k) in predictions]
return json.dumps(clean_predictions)
# In[ ]:
training_set_mobilenet_features = []
for i in list(range(1, 401)) + list(range(501, 901)):
img_path = 'test8/%06d.png' % i
# print(img_path)
img = image.load_img(img_path, target_size=(224, 224)) # 读取图片
#plt.imshow(img)
#plt.show()# 显示图片
x = image.img_to_array(img) # 图片转换为ndarray
x = np.expand_dims(x, axis=0)
x = preprocess_input(x) # 数组转换为mobilenet输入格式
training_set_mobilenet_features.append(
model_mobilenet.predict(x).reshape((7 * 7 * 1024, ))) # 计算该张图片的特征
# In[ ]:
training_set_mobilenet_features_ndarray = np.vstack(
training_set_mobilenet_features) # 转换为ndarray
# In[ ]:
training_set_label = np.array([1.0 if i < 400 else 0.0 for i in range(800)])
# In[ ]:
test_set_mobilenet_features = []
for i in list(range(401, 501)) + list(range(901, 1001)):
img_path = 'test8/%06d.png' % i
for _ in camera.capture_continuous(
stream, format='jpeg', use_video_port=True): #start to capture frames from the preview
stream.seek(0) #begin from the first position of the buffer
img_ov = Image.new("RGBA", (CAMERA_WIDTH,CAMERA_HEIGHT)) #create the overlay using the same resolution of the camera
draw = ImageDraw.Draw(img_ov) #we are going to write on this overlay
draw.font = ImageFont.truetype("/usr/share/fonts/truetype/freefont/FreeSans.ttf", 20)
draw.rectangle([5,5,CAMERA_WIDTH-5,CAMERA_HEIGHT-5], fill=None, outline="red") #a purely decorative choice
img = Image.open(stream).convert('RGB').resize((224, 224)) #we modify the frame to give it as input to the model
#we then give the frame to the model and calculate the inference time
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
start_time = time.monotonic()
preds = model.predict(x)
elapsed_ms = (time.monotonic() - start_time) * 1000
preds= decode_predictions(preds)
# we write the first two output probabilities onto the overlay
draw.multiline_text((8,40), str(preds[0][0][1])+"\n"+str('%.1f' % ((preds[0][0][2])*100))+
" % \n", fill=(255,0,0,255), font=None, anchor=None, spacing=5, align="left")
draw.multiline_text((8,85), str(preds[0][1][1])+"\n"+str('%.1f' % ((preds[0][1][2])*100))+
" % \n", fill=(255,0,0,255), font=None, anchor=None, spacing=5, align="left")
draw.text((8,8),str('%.1f ms' % (elapsed_ms)), fill="red", font=None, anchor=None)
if not overlay_renderer:
"""
If overlay layer is not created yet, get a new one. Layer
parameter must have 3 or higher number because the original
preview layer has a # of 2 and a layer with smaller number will
camera = picamera.PiCamera()
camera.resolution = (224, 224)
model = MobileNet(weights='imagenet')
devnull = open('os.devnull', 'w')
i = 1
while True:
camera.capture("img.jpg")
img = image.load_img("img.jpg", target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
pred = model.predict(x)
word = decode_predictions(pred)[0][0][1].replace('_', ' ')
word2 = decode_predictions(pred)[0][1][1].replace('_', ' ')
speak = "Okay I got it. I believe I'm looking at some kind of {} or {}".format(
word, word2)
print(speak)
os.rename('img.jpg',
'outputs/mobilenet/{}-{}-{}.jpg'.format(i, word, word2))
subprocess.run([
'pico2wave', '-w', 'outputs/mobilenet/{}-result.wav'.format(i),
"{}".format(speak)
],
stdout=devnull,
stderr=subprocess.STDOUT)
subprocess.call(['aplay', 'outputs/mobilenet/{}-result.wav'.format(i)])
print()
return result
# load the MobileNet model
model_features = MobileNet(input_shape=(IMG_WIDTH, IMG_HEIGHT, 3),
include_top=False,
weights='imagenet')
img_path = 'test.png'
img = image.load_img(img_path, target_size=(224, 224))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
# get the bottleneck features from the mobilenet
bottleneck_features = model_features.predict(x)
# load our classfier model
model_classification = load_model(TOP_MODEL_PATH)
# use the bottleneck features to get the final classification
class_predicted = model_classification.predict(bottleneck_features)
#print('Predicted:', decode_predictions(preds, top=3)[0])
preds = decode_lights_predictions(class_predicted)
print(preds)
"""
orig = cv2.imread(img_path)
from picamera.array import PiRGBArray
from picamera import PiCamera
model = MobileNet(weights='imagenet')
print "Load model successfully"
camera = PiCamera()
camera.resolution = (600, 600)
camera.framerate = 32
cap = PiRGBArray(camera, size=(600, 600))
print "Start video stream"
for i in camera.capture_continuous(cap, format='bgr', use_video_port=True):
frame = i.array
image = cv2.resize(frame, (224, 224))
img = image.img_to_array(img)
img = np.expand_dims(img, axis=0)
img = preprocess_input(img)
preds = model.predict(img)
cv2.putText(frame,
'Predicted: ' + str(decode_predictions(preds, top=1)[0]),
(10, 25), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 255, 0), 2)
cv2.imshow("Frame", frame)
key = cv2.waitKey(1) & 0xFF
cap.truncate(0)
if key == ord('q'):
break
img_mobilenet = load_img(img_mobilenet_pth, target_size=(
224, 224)) # image size can be calibrated with target_size parameter
img_mobilenet
img_mobilenet = img_to_array(img_mobilenet)
print(img_mobilenet.shape)
img_mobilenet = np.expand_dims(img_mobilenet, axis=0)
print(img_mobilenet.shape)
img_mobilenet = preprocess_input(
img_mobilenet) # preprocess image with preprocess_input function
print(img_mobilenet.shape)
## prediction with preprocessing
pred_mobilenet = model_mobilenet.predict(img_mobilenet)
# print(pred_class)
n = 10
top_n_mobilenet = decode_predictions(pred_mobilenet, top=n)
for b in top_n_mobilenet[0]:
print(b)
##4
#using VGG16 MODEL
from keras.applications.vgg16 import VGG16
model_vgg16 = VGG16()
#load image
class AppearanceFeaturesExtractor(object):
"""
"""
def __init__(self,
model_type="MobileNet",
weights="imagenet",
input_shape=(128, 128, 3)):
"""
"""
model_types = ["MobileNet", "MobileNetV2", "VGG16", "ResNet50"]
weights_types = ["imagenet", "random"]
if weights not in weights_types:
raise ValueError(
"Invalid weights. Should be one of: {}".format(weights_types))
if model_type not in model_types:
raise ValueError(
"Invalid model type. Should be one of: {}".format(model_types))
assert input_shape[0] > 32
assert input_shape[1] > 32
self.input_shape = input_shape
if model_type == "MobileNet":
self.model = MobileNet(weights=weights,
include_top=False,
pooling='avg',
input_shape=input_shape)
if model_type == "MobileNetV2":
self.model = MobileNetV2(weights=weights,
include_top=False,
pooling='avg',
input_shape=input_shape)
if model_type == "VGG16":
self.model = VGG16(weights=weights,
include_top=False,
pooling='avg',
input_shape=input_shape)
if model_type == "ResNet50":
self.model = ResNet50(weights=weights,
include_top=False,
pooling='avg',
input_shape=input_shape)
def extract(self, rgb_image, track=None):
"""
"""
if track is not None:
x = track.bbox.xmin
y = track.bbox.ymin
w = track.bbox.width()
h = track.bbox.height()
crop_image = rgb_image[y:y + h, x:x + w]
else:
crop_image = rgb_image
image_resized = cv2.resize(crop_image,
(self.input_shape[0], self.input_shape[1]))
x = image.img_to_array(image_resized)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)
features = self.model.predict(x)
return np.array(features).flatten()
def mobile_net(pred_cats,image_path,pickle_path,query_path):
model = MobileNet(weights='imagenet', include_top=False)
#Reading Image Names from Predicted Categories
image_names=[]
for i in range(0,len(pred_cats)) :
# Give the path of the semi trained dataset
path = image_path +"/"+ pred_cats[i]
for img_path in os.listdir(path):
if('.DS_Store' not in os.path.join(path,img_path)):
image_names.append(pred_cats[i]+"/"+img_path)
start = time.time()
#print (image_names)
# Give the MobileNet pickle file path
for i in range(3):
pickled_db_path = pickle_path + pred_cats[i] + ".pck"
with open(pickled_db_path, 'rb') as fp:
temp = pickle.load(fp)
fp.close()
if i == 0:
mnet_loaded = temp
else:
mnet_loaded = np.append(mnet_loaded,temp,axis = 0)
print(mnet_loaded.shape)
#Extract features of Query Image
imgq = image.load_img(query_path, target_size=(224, 224))
img_dataq = image.img_to_array(imgq)
img_dataq = np.expand_dims(img_dataq, axis=0)
img_dataq = preprocess_input(img_dataq)
mnet_feature_query = model.predict(img_dataq)
mnet_feature_np_query = np.array(mnet_feature_query)
mnet_feature_np_query = mnet_feature_np_query.flatten()
#Calculate Cosine Similarity
similarity = []
count = 0
for i in mnet_loaded:
count = count+1
d = distance.cosine(i,mnet_feature_np_query)
sim = 1-d
similarity.append((sim,image_names[count-1]))
# To display Image
def show_img(path):
img =imread(path, mode="RGB")
plt.imshow(img)
plt.show()
# Picking top 6 image similarities
des_similarity = sorted(similarity,reverse=True)
des_similarity = des_similarity[:6]
print(des_similarity)
# Give the Semi_images path
print ('Query image ==========================================')
show_img(query_path)
print ('Result images ========================================')
for i in range(len(des_similarity)):
print("similarity",des_similarity[i][0])
result_image_path = image_path+ "/"+des_similarity[i][1]
if(os.path.isfile(result_image_path)):
show_img(result_image_path)
end = time.time()
total_time_main = end - start
print("total_time_main", total_time_main)
# capture frames from the camera
for frame in camera.capture_continuous(rawCapture,
format="bgr",
use_video_port=True):
# grab the raw NumPy array representing the image, then initialize the timestamp
# and occupied/unoccupied text
image = frame.array
# machine learning
# resize to VGG16 size(224, 224)
img = Image.fromarray(np.uint8(image))
img = img.resize((224, 224))
x = img
pred_data = np.expand_dims(x, axis=0)
preds = model.predict(preprocess_input(pred_data))
results = decode_predictions(preds, top=1)[0]
for result in results:
# print(result)
label = result[1]
accu = str(result[2])
#gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# show the frame
new_label = label + accu
image = cv2.putText(image, new_label, (10, 25), cv2.FONT_HERSHEY_SIMPLEX,
0.7, (0, 255, 0), 2)
cv2.imshow("Frame", image)
key = cv2.waitKey(1) & 0xFF
# ピカ処理
y_train = to_categorical(y_train) y_test = to_categorical(y_test) # In[11]: print(x_train.shape, y_train.shape) print(x_test.shape, y_test.shape) # In[12]: from keras.applications.mobilenet import MobileNet MobileNet_extractor = MobileNet(weights='imagenet', include_top=False) # In[13]: train_features = MobileNet_extractor.predict(x_train) test_features = MobileNet_extractor.predict(x_test) # In[14]: print(train_features.shape, y_train.shape) print(test_features.shape, y_test.shape) # In[15]: train_features = train_features.reshape(train_features.shape[0], -1) test_features = test_features.reshape(test_features.shape[0], -1) # In[16]: print(train_features.shape, y_train.shape)
"""
name = os.listdir('/content/drive/My Drive/Colab Notebooks/Large_data/Test')
test_images = []
for i in range(len(name)):
img = cv2.imread('/content/drive/My Drive/Colab Notebooks/Large_data/Test/' + name[i],0)
img = cv2.resize(img,(64,64))
test_images.append(img)
#print('Test' + str(i))
test_images = np.array(test_images)
test_images = test_images/255.0
test_images = np.expand_dims(test_images,3)
final = model.predict(test_images)
print(final)
pred = []
for i in final:
pred.append(np.argmax(i))
print(pred)
"""
Plot metrics
"""
print(example.history.keys())
plt.plot(example.history['acc'])
plt.plot(example.history['val_acc'])
# In[ ]:
training_set_vgg16_features = []
for i in list(range(1, 401)) + list(range(501, 901)):
img_path = 'test8/%06d.png' % i
# print(img_path)
img = image.load_img(img_path, target_size=(224, 224)) # 读取图片
#plt.imshow(img)
#plt.show()# 显示图片
x = image.img_to_array(img) # 图片转换为ndarray
x = np.expand_dims(x, axis=0)
x = preprocess_input(x) # 数组转换为vgg16输入格式
training_set_vgg16_features.append(
model_vgg16.predict(x).reshape((7 * 7 * 512, ))) # 计算该张图片的特征
# In[ ]:
training_set_vgg16_features_ndarray = np.vstack(
training_set_vgg16_features) # 转换为ndarray
# In[ ]:
training_set_label = np.array([1.0 if i < 400 else 0.0 for i in range(800)])
# # 读取测试集图片计算特征
# In[ ]:
test_set_vgg16_features = []
# loop over the frames from the video stream
while True:
# grab the frame from the threaded video stream and resize it
# to have a maximum width of 400 pixels
frame = vs.read()
frame = imutils.resize(frame, width=400)
# prepare the image to be classified by our deep learning network
image = cv2.resize(frame, (224, 224))
image = image.astype("float") / 255.0
image = img_to_array(image)
image = np.expand_dims(image, axis=0)
# classify the input image and initialize the label and
# probability of the prediction
preds = model.predict(preprocess_input(image))
results = decode_predictions(preds, top=1)[0]
for result in results:
# print(result)
label = result[1]
proba = str(result[2])
# otherwise, reset the total number of consecutive frames and the
# santa alarm
# build the label and draw it on the frame
label = "{0}: {1}%".format(label, proba * 100)
frame = cv2.putText(frame, label, (10, 25), cv2.FONT_HERSHEY_SIMPLEX, 0.7,
(0, 255, 0), 2)
# show the output frame
openNames = []
files = os.listdir(test_file_dir)
for x in files:
if os.path.isfile(test_file_dir + x):
openNames.append(test_file_dir + str(x))
image = load_img(openNames[len(openNames) - 1], target_size=(224, 224))
print("loaded image #", len(openNames) - 1)
# convert the image pixels to a numpy array
image = img_to_array(image)
# reshape data for the model
image = image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))
# prepare the image for the VGG model
image = preprocess_input(image)
# predict the probability across all output classes
yhat = model.predict(image)
# convert the probabilities to class labels
labelPredict = decode_predictions(yhat)
# retrieve the most likely result, e.g. highest probability
label = labelPredict[0][0]
label2 = labelPredict[0][1]
label3 = labelPredict[0][2]
# print the classification
print('%s (%.2f%%)' % (label[1], label[2] * 100))
print('%s (%.2f%%)' % (label2[1], label2[2] * 100))
print('%s (%.2f%%)' % (label3[1], label3[2] * 100))
openFile = open("C://Users//mitch//Desktop//ImgText//guess.txt", 'w')
outString = str(label[1]) + "," + str(label[2])
print(outString)
print(outString, file=openFile)
openFile.close()
# Load our model
# model = densenet169_model(img_rows=img_rows, img_cols=img_cols, color_type=channel, num_classes=num_classes)
# load keras model
model = MobileNet(weights=None, classes=10)
sgd = SGD(lr=1e-3, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(optimizer=sgd,
loss='categorical_crossentropy',
metrics=['accuracy'])
model.summary()
# Start Fine-tuning
model.fit(
X_train,
Y_train,
batch_size=batch_size,
epochs=nb_epoch,
shuffle=True,
verbose=1,
validation_data=(X_valid, Y_valid),
)
# Make predictions
predictions_valid = model.predict(X_valid,
batch_size=batch_size,
verbose=1)
# Cross-entropy loss score
score = log_loss(Y_valid, predictions_valid)
