def load_data(path):
print("[INFO] loading images...")
data = []
labels = []
# grab the image paths and randomly shuffle them
imagePaths = sorted(list(paths.list_images(path)))
random.seed(42)
random.shuffle(imagePaths)
# loop over the input images
for imagePath in imagePaths:
# load the image, pre-process it, and store it in the data list
image = cv2.imread(imagePath)
image = cv2.resize(image, (norm_size, norm_size))
image = img_to_array(image)
data.append(image)
# extract the class label from the image path and update the
# labels list
label = int(imagePath.split(os.path.sep)[-2])
labels.append(label)
# scale the raw pixel intensities to the range [0, 1]
data = np.array(data, dtype="float") / 255.0
labels = np.array(labels)
# convert the labels from integers to vectors
labels = to_categorical(labels, num_classes=CLASS_NUM)
return data,labels
def extract():
imagepaths = sorted(paths.list_images('training_data/'))
labels = []
data = []
for imagepath in imagepaths:
label = imagepath[imagepath.rfind('/') + 1:].split('-')[0]
image = cv2.imread(imagepath)
features = describe_haralick_stats(image)
labels.append(label)
data.append(features)
datafile = "models/features.pkl"
td_file = open(datafile, 'wb')
pickle.dump(data, td_file)
td_file.close()
labelfile = "models/labels.pkl"
ld_file = open(labelfile, 'wb')
pickle.dump(labels, ld_file)
ld_file.close()
return
def get_training_images(training_dir):
training_images = dict()
for image_path in paths.list_images(training_dir):
leaf_sort = image_path.split("/")[-2]
if leaf_sort in training_images:
training_images[leaf_sort].append(image_path)
else:
training_images[leaf_sort] = [image_path]
return training_images
def process_dir(dataset_path, output):
files = paths.list_images(dataset_path)
# for fname in files:
f = open(output, 'a+')
total = 0
# f.write("{}\n".format(encodings.base64_encode_image(face)))
total += 1
print("[INFO] wrote {} frames to file".format(total))
f.close()
def get_train_data(path,img_cols,img_rows,num_imgs=-1):
print path
i=1
x_img=[]
y_labels=[]
if num_imgs!="-1":
for imagePath in paths.list_images(path):
i+=1
print(imagePath,str(i))
img=cv2.imread(imagePath)
imagePath=imagePath.split("/")
y_labels.append(int(imagePath[1].replace("c","")))#once we split, lists are like this
#[train,c0,img3012341.jpg]
#we want only integers for CNN model
res = cv2.resize(img, (img_cols, img_rows))
res=res.transpose(2,0,1)#Size is now 3x32x32
x_img.append(res)
if i==num_imgs:
return x_img,y_labels
print i
else:
print("HOLAAA")
for imagePath in paths.list_images(path):
i+=1
print(imagePath,str(i))
img=cv2.imread(imagePath)
imagePath=imagePath.split("/")
y_labels.append(int(imagePath[1].replace("c","")))#once we split, lists are like this
#[train,c0,img3012341.jpg]
#we want only integers for CNN model
res = cv2.resize(img, (img_cols, img_rows))
res=res.transpose(2,0,1)#Size is now 3x32x32
x_img.append(res)
return x_img,y_labels
def create(path):
images = list(list_images(path))
outfile = '/tmp/out.avi'
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = None
fps = 3
print(len(images))
for img in sorted(images):
image = cv2.imread(img)
if out is None:
h, w = image.shape[:2]
print(h, w)
out = cv2.VideoWriter(outfile, fourcc, fps, (w, h))
out.write(image)
def create_gif(inputPath, outputPath, delay, finalDelay, loop):
# grab all image paths in the input directory
imagePaths = sorted(list(paths.list_images(inputPath)))
# remove the last image path in the list
lastPath = imagePaths[-1]
imagePaths = imagePaths[:-1]
# construct the image magick 'convert' command that will be used
# generate our output GIF, giving a larger delay to the final
# frame (if so desired)
cmd = "convert -delay {} {} -delay {} {} -loop {} {}".format(
delay, " ".join(imagePaths), finalDelay, lastPath, loop,
outputPath)
os.system(cmd)
def read_repre(repre_path):
# f = open(repre_path, 'r')
# repres = dict()
# for l in f:
# elems = l.strip().split(' ')
# repres[elems[0]] = cv2.imread(elems[1])
# f.close()
# return repres
files = paths.list_images(repre_path)
repres = {}
for f in files:
name = f.split('/')[-1]
name = name[:name.rfind('.')]
img = cv2.imread(f)
repres[name] = img
return repres
def hyperparameters_tuning(img):
scaleFactor_list = [1.05, 1.1, 1.2]
minNeighbors_list = [5, 8, 11]
minSize_list = [60, 80, 100]
it = len(scaleFactor_list) * len(minNeighbors_list) * len(minSize_list)
curr_it = 0
res_file = open('/home/tomas/Dropbox/Data/celeb_dataset/hyperparameters_tuning.txt' 'w')
for sf in scaleFactor_list:
for mn in minNeighbors_list:
for ms in minSize_list:
missed = 0
extra = 0
ok = 0
n_files = 0
curr_it += 1
files = paths.list_images(dataset_path)
# for fname in tqdm(files):
for fname in files:
n_files += 1
image = cv2.imread(fname)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# gray = imutils.resize(gray, width=width)
im_face, n_faces, bbox, = detect_face(gray, fname, scaleFactor=sf, minNeighbors=mn, minSize=ms,
save_face=save_face)
if n_faces == 0:
missed += 1
elif n_faces > 1:
extra += n_faces - 1
else:
ok += 1
# print 'total # of images: ', n_files
print 'it #%i/%i -----------------------------------------------' % (curr_it, it)
print 'params: sf=%.2f, mn=%i, ms=%i' % (sf, mn, ms)
print 'ok: %i/%i, missed: %i, extra: %i' % (ok, n_files, missed, extra)
print '\n'
res_file.write('it #%i/%i -----------------------------------------------' % (curr_it, it))
res_file.write('params: sf=%.2f, mn=%i, ms=%i' % (sf, mn, ms))
res_file.write('ok: %i/%i, missed: %i, extra: %i' % (ok, n_files, missed, extra))
res_file.write('\n')
def create_base64_files(dataset_path, base64_path, face_cascade):
if not os.path.exists(base64_path):
os.mkdir(base64_path)
fd = FaceDetector(face_cascade)
print("[INFO] Extracting faces from dataset...")
files = paths.list_images(dataset_path)
data_dict = {}
for f in files:
# name = f.split('/')[-2][:f.split('/')[-1].rfind('.')]
# label = name[:name.rfind('_')]
label = f.split('/')[-2]
img = cv2.imread(f, 0)
img = imutils.resize(img, width=500)
# faceRects = fd.detect(img, scaleFactor=1.1, minNeighbors=9, minSize=(100, 100))
faceRects = fd.detect(img, scaleFactor=1.1, minNeighbors=9, minSize=(40, 40))
if len(faceRects) > 0:
(x, y, w, h) = max(faceRects, key=lambda b: (b[2] * b[3]))
face = img[y:y + h, x:x + w].copy(order="C")
else:
# cv2.imshow('No face detected', img)
# cv2.waitKey(0)
# raise ValueError('No face deteted.')
continue
if data_dict.has_key(label):
data_dict[label].append(face)
else:
data_dict[label] = list((face,))
# for fname in files:
for k, v in data_dict.items():
filename = k + '.txt'
filepath = os.path.join(base64_path, filename)
f = open(filepath, 'a+')
total = 0
for face in v:
f.write("{}\n".format(encodings.base64_encode_image(face)))
total += 1
print("[INFO] wrote {} frames to file {}".format(total, filename))
f.close()
def detect_batch(dataset_path, save_face=False):
scaleFactor = 1.1
minNeighbors = 8
minSize = 60
missed = 0
extra = 0
ok = 0
n_files = 0
files = paths.list_images(dataset_path)
faces = []
names = []
# files = ['/home/tomas/Dropbox/Data/celeb_dataset/natalie_portman/natalie_portman_07.jpg', ]
for fname in files:
n_files += 1
image = cv2.imread(fname)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
im_face, n_faces, bbox = detect_face(gray, fname, scaleFactor=scaleFactor, minNeighbors=minNeighbors,
minSize=minSize, save_face=save_face)
if n_faces == 0:
missed += 1
print 'missed:', fname.split('/')[-1]
elif n_faces > 1:
extra += n_faces - 1
print 'extra:', fname.split('/')[-1]
else:
ok += 1
# imgs.append(im_face)
bbox = bbox[0]
# bboxes.append(bbox)
face = gray[bbox[1]:bbox[1] + bbox[3], bbox[0]:bbox[0] + bbox[2]]
faces.append(face)
name = fname.split('/')[-1].split('.')[0]
names.append(name)
# print 'total # of images: ', n_files
print 'ok: %i/%i, missed: %i, extra: %i' % (ok, n_files, missed, extra)
return faces, names
def load_images(if_grayscale=False, if_normalize_images=IF_NORMALIZE_IMAGES):
print('Loading image data...')
images = []
image_labels = []
# For every class, for every image in its images dataset, accumulate into list of features & labels
for (i, clazz) in enumerate(CLASSES):
image_paths = list(paths.list_images(clazz.get('image_dir_path')))
image_paths = image_paths[0:IMAGE_COUNT_PER_CLASS] if IMAGE_COUNT_PER_CLASS else image_paths
for image_path in image_paths:
image = cv2.imread(
image_path,
cv2.IMREAD_GRAYSCALE if if_grayscale else cv2.IMREAD_COLOR,
)
if if_normalize_images:
image = cv2.normalize(image, None, 0, 255, cv2.NORM_MINMAX)
images.append(image)
image_labels.append(i)
return (images, image_labels)
def extract(conf):
# grab the set of ground-truth images and select a percentage of them for training
trnPaths = list(paths.list_images(conf["pos_raw_ph"]))
assert len(trnPaths)!=0
#
if not os.path.isdir(conf['pos_feat_ph']):
os.makedirs(conf['pos_feat_ph'])
# setup the progress bar
widgets = ["Extracting: ", progressbar.Percentage(), " ", progressbar.Bar(), " ", progressbar.ETA()]
pbar = progressbar.ProgressBar(maxval=len(trnPaths), widgets=widgets).start()
for (i, imgPath) in enumerate(trnPaths):
# img = imread(imgPath)
# resize to the training resolution (wid,height)
img = auto_resized(imread(imgPath, as_grey=True),conf['sliding_size'])
fd=hog(img, conf['orientations'], conf['pixels_per_cell'],
conf['cells_per_block'], conf['visualize'], conf['normalize'])
fd_name = os.path.split(imgPath)[1].split(".")[0] + ".feat"
fd_path = os.path.join(conf['pos_feat_ph'], fd_name)
joblib.dump(fd, fd_path)
pbar.update(i)
pbar.finish()
print '[*] Finished Pos.'
def data2files(data_dir, face_cascade, res_file_path, label_dict_path):
# initializing face detector
fd = FaceDetector(face_cascade)
dataset = []
files = paths.list_images(data_dir)
file_dict = process_files(files)
for key in file_dict.keys():
print 'Processing %s ...' % key,
celeb_files = file_dict[key]
n_files = len(celeb_files)
ok = 0
for f in celeb_files:
item = create_data_item(f, fd)
if item is not None:
dataset.append(item)
ok += 1
print 'done, face found: %i/%i (%i%%)' % (ok, n_files, int(100 * ok / n_files))
print 'Saving data in pickle format...',
# data_fname = os.path.join(res_dir, 'data.npz')
data_file = gzip.open(res_file_path, 'wb', compresslevel=1)
pickle.dump(dataset, data_file)
# np.save(data_file, np.array(data_cubes_resh))
data_file.close()
print 'done'
print 'Saving label dictionary ...'
label_dict = {}
names = file_dict.keys()
for i, n in enumerate(names):
label_dict[n] = i
labels_file = gzip.open(label_dict_path, 'wb', compresslevel=1)
pickle.dump(label_dict, labels_file)
labels_file.close()
print 'done'
ap.add_argument("-d", "--dataset", required=True, help="path to input dataset")
#ap.add_argument("-a", "--augment", type=int, default=-1,
# help="whether or not 'on the fly' data augmentation should be used")
ap.add_argument("-p",
"--plot",
type=str,
default="plot.png",
help="path to output loss/accuracy plot")
args = vars(ap.parse_args())
# # Using Resnet50 and initialised with weights of imagenet
# ## images in smear 2005 are resized to 224 x224
# grab the list of images in our dataset directory, then initialize
# the list of data (i.e., images) and class images
print("[INFO] loading images...")
imagePaths = list(paths.list_images(args["dataset"]))
data = []
labels = []
# loop over the image paths
for imagePath in imagePaths:
# extract the class label from the filename, load the image, and
# resize it to be a fixed 64x64 pixels, ignoring aspect ratio
label = imagePath.split(os.path.sep)[-2]
image = cv2.imread(imagePath, 1)
image = cv2.resize(image, (224, 224))
# update the data and labels lists, respectively
data.append(image)
labels.append(label)
"--images",
required=True,
help="path to input directory of images")
ap.add_argument(
"-t",
"--threshold",
type=float,
default=100.0,
help="focus measures that fall below this value will be considered 'blurry'"
)
args = vars(ap.parse_args())
outputfolder_path = "/content/drive/MyDrive/checking-blur/sharpimages/"
i = 0
# loop over the input images
for imagePath in paths.list_images(args["images"]):
# load the image, convert it to grayscale, and compute the
# focus measure of the image using the Variance of Laplacian
# method
name = path.splitext(os.path.basename(imagePath))[0]
image = cv2.imread(imagePath)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
fm = variance_of_laplacian(gray)
text = "Not Blurry"
# if the focus measure is less than the supplied threshold,
# then the image should be considered "blurry"
if fm < args["threshold"]:
text = "Blurry"
# show the image
cv2.putText(image, "{}: {:.2f}".format(text, fm), (10, 30),
cv2.FONT_HERSHEY_SIMPLEX, 0.8, (0, 0, 255), 3)
# USAGE
# python build_dogs_vs_cats.py
from config import dogs_and_cats_config as config
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from pyimagesearch.preprocessing import AspectAwarePreprocessor
from pyimagesearch.io import HDF5DatasetWriter
from imutils import paths
import numpy as np
import progressbar
import json
import cv2
import os
# grab the paths to the images
train_paths = list(paths.list_images(config.IMAGES_PATH))
train_labels = [p.split(os.path.sep)[-1].split('.')[0] for p in train_paths]
le = LabelEncoder()
train_labels = le.fit_transform(train_labels)
# perform stratified sampling from the training set to build the testing split from the training data
train_paths, test_paths, train_labels, test_labels = train_test_split(
train_paths,
train_labels,
test_size=config.NUM_TEST_IMAGES,
stratify=train_labels,
random_state=42)
train_paths, val_paths, train_labels, val_labels = train_test_split(
train_paths,
train_labels,
test_size=config.NUM_VAL_IMAGES,
TARGET_IMAGE_WIDTH = 28
TARGET_IMAGE_HEIGHT = 28
LR = 0.001 # 学习率
BATCH_SIZE = 64
EPOCHS = 40
################################################
# 第一部分：数据预处理
# initialize the image preprocessor and datasetloader
sp = SimplePreprocessor(TARGET_IMAGE_WIDTH, TARGET_IMAGE_HEIGHT)
sdl = SimpleDatasetLoader(preprocessors=[sp])
# Load images
print("[INFO] 导入图像...")
image_paths = list(paths.list_images(dataset_path)) # path includedd
print(paths.list_images(dataset_path))
(X, y) = sdl.load(image_paths, verbose=500, grayscale=True)
# Flatten (reshape the data matrix)
# convert from (13164,TARGET_IMAGE_WIDTH,TARGET_IMAGE_HEIGHT)
#into (13164,TARGET_IMAGE_WIDTH*TARGET_IMAGE_HEIGHT)
X = X.reshape((X.shape[0], TARGET_IMAGE_WIDTH * TARGET_IMAGE_HEIGHT))
X = X.astype("float") / 255.0 # 特征缩放，是非常重要的步骤
# Show some information on memory consumption of the images
print("[INFO] features matrix: {:.1f}MB".format(X.nbytes / (1024 * 1024.0)))
# Label encoder
le = LabelEncoder()
y = to_categorical(le.fit_transform(y), 2)
from tensorflow.keras.models import load_model
import config
from imutils import paths
import numpy as np
import imutils
import random
import cv2
import os
# load the trained model from disk
print("[INFO] loading model...")
model = load_model(config.MODEL_PATH)
# grab the paths to the fire and non-fire images, respectively
print("[INFO] predicting...")
firePaths = list(paths.list_images(config.FIRE_PATH))
nonFirePaths = list(paths.list_images(config.NON_FIRE_PATH))
# combine the two image path lists, randomly shuffle them, and sample
# them
imagePaths = firePaths + nonFirePaths
random.shuffle(imagePaths)
imagePaths = imagePaths[:config.SAMPLE_SIZE]
# loop over the sampled image paths
for (i, imagePath) in enumerate(imagePaths):
# load the image and clone it
image = cv2.imread(imagePath)
output = image.copy()
# resize the input image to be a fixed 128x128 pixels, ignoring
def predictImages(modelArg, labelsArg, imagePathArg, num_classesArg,
min_confidenceArg, image_displayArg, pred_stagesArg):
# initialize the model
model = tf.Graph()
# create a context manager that makes this model the default one for
# execution
with model.as_default():
# initialize the graph definition
graphDef = tf.GraphDef()
# load the graph from disk
with tf.gfile.GFile(modelArg, "rb") as f:
serializedGraph = f.read()
graphDef.ParseFromString(serializedGraph)
tf.import_graph_def(graphDef, name="")
# load the class labels from disk
labelMap = label_map_util.load_labelmap(labelsArg)
categories = label_map_util.convert_label_map_to_categories(
labelMap, max_num_classes=num_classesArg, use_display_name=True)
categoryIdx = label_map_util.create_category_index(categories)
# create a plateFinder
plateFinder = PlateFinder(min_confidenceArg,
categoryIdx,
rejectPlates=False,
charIOUMax=0.3)
# create plate displayer
plateDisplay = PlateDisplay()
# create a session to perform inference
with model.as_default():
with tf.Session(graph=model) as sess:
# create a predicter, used to predict plates and chars
predicter = Predicter(model, sess, categoryIdx)
imagePaths = paths.list_images(imagePathArg)
frameCnt = 0
start_time = time.time()
platesReturn = []
numPlates = 0
# Loop over all the images
for imagePath in imagePaths:
frameCnt += 1
# load the image from disk
print("[INFO] Loading image \"{}\"".format(imagePath))
image = cv2.imread(imagePath)
(H, W) = image.shape[:2]
# If prediction stages == 2, then perform prediction on full image, find the plates, crop the plates from the image,
# and then perform prediction on the plate images
if pred_stagesArg == 2:
# Perform inference on the full image, and then select only the plate boxes
boxes, scores, labels = predicter.predictPlates(
image, preprocess=True)
licensePlateFound_pred, plateBoxes_pred, plateScores_pred = plateFinder.findPlatesOnly(
boxes, scores, labels)
# loop over the plate boxes, find the chars inside the plate boxes,
# and then scrub the chars with 'processPlates', resulting in a list of final plateBoxes, char texts, char boxes, char scores and complete plate scores
plates = []
for plateBox in plateBoxes_pred:
boxes, scores, labels = predicter.predictChars(
image, plateBox)
chars = plateFinder.findCharsOnly(
boxes, scores, labels, plateBox, image.shape[0],
image.shape[1])
if len(chars) > 0:
plates.append(chars)
else:
plates.append(None)
plateBoxes_pred, charTexts_pred, charBoxes_pred, charScores_pred, plateAverageScores_pred = plateFinder.processPlates(
plates, plateBoxes_pred, plateScores_pred)
# If prediction stages == 1, then predict the plates and characters in one pass
elif pred_stagesArg == 1:
# Perform inference on the full image, and then find the plate text associated with each plate
boxes, scores, labels = predicter.predictPlates(
image, preprocess=False)
licensePlateFound_pred, plateBoxes_pred, charTexts_pred, charBoxes_pred, charScores_pred, plateScores_pred = plateFinder.findPlates(
boxes, scores, labels)
else:
print(
"[ERROR] --pred_stages {}. The number of prediction stages must be either 1 or 2"
.format(pred_stagesArg))
quit()
# Print plate text
for charText in charTexts_pred:
print(" Found: ", charText)
# Display the full image with predicted plates and chars
if image_displayArg == True:
imageLabelled = plateDisplay.labelImage(
image, plateBoxes_pred, charBoxes_pred, charTexts_pred)
cv2.imshow("Labelled Image", imageLabelled)
cv2.waitKey(0)
imageResults = []
for i, plateBox in enumerate(plateBoxes_pred):
imageResults.append({
'plateText':
charTexts_pred[i],
'plateBoxLoc':
list(plateBox),
'charBoxLocs':
list([list(x) for x in charBoxes_pred[i]])
})
numPlates += 1
platesReturn.append({
'imagePath': imagePath.split("/")[-1],
'imageResults': imageResults
})
# print some performance statistics
curTime = time.time()
processingTime = curTime - start_time
fps = frameCnt / processingTime
print(
"[INFO] Processed {} frames in {:.2f} seconds. Frame rate: {:.2f} Hz"
.format(frameCnt, processingTime, fps))
#results = results.encode('utf-8')
return {
"processingTime": processingTime,
"numPlates": numPlates,
"numImages": len(platesReturn),
"images": platesReturn
}
def train_network(num_epochs=30, LETTER_IMAGES_FOLDER="images"):
"""
Trains a convolutional neural network to our training set and outputs the
model to a local file to be read and used
"""
MODEL_FILENAME = "model.hdf5"
MODEL_LABELS_FILENAME = "model_labels.dat"
# initialize the data and labels
data = []
labels = []
# loop over the input images
for image_file in paths.list_images(LETTER_IMAGES_FOLDER):
# Load the image and convert it to grayscale
image = cv2.imread(image_file, 0)
if image is None:
# the filename was probably too long
print(
"WARNING: We couldn't open this image, its filename is probably too long:"
)
print(image_file)
continue
# threshold
_, image = cv2.threshold(image, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
image = crop_letter(image)
# Resize the letter so it fits in a 20x20 pixel box
image = resize_to_fit(image, 20, 20)
# Add a third channel dimension to the image to make Keras happy
image = np.expand_dims(image, axis=2)
# Grab the name of the letter based on the folder it was in
label = image_file.split(os.path.sep)[-1].replace('.png', '')
# Add the letter image and it's label to our training data
data.append(image)
labels.append(label)
# scale the raw pixel intensities to the range [0, 1] (this improves training)
data = np.array(data, dtype="float") / 255.0
labels = np.array(labels)
# Split the training data into separate train and test sets
(X_train, X_test, Y_train, Y_test) = train_test_split(data,
labels,
test_size=0.25,
random_state=0)
# Convert the labels (letters) into one-hot encodings that Keras can work with
lb = LabelBinarizer().fit(Y_train)
Y_train = lb.transform(Y_train)
Y_test = lb.transform(Y_test)
# Save the mapping from labels to one-hot encodings.
# We'll need this later when we use the model to decode what it's predictions mean
with open(MODEL_LABELS_FILENAME, "wb") as f:
pickle.dump(lb, f)
# Build the neural network!
model = Sequential()
# First convolutional layer with max pooling
model.add(
Conv2D(20, (5, 5),
padding="same",
input_shape=(20, 20, 1),
activation="relu"))
model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))
# Second convolutional layer with max pooling
model.add(Conv2D(50, (5, 5), padding="same", activation="relu"))
model.add(MaxPooling2D(pool_size=(2, 2), strides=(2, 2)))
# Hidden layer with 500 nodes
model.add(Flatten())
model.add(Dense(500, activation="relu"))
# Output layer with 70 nodes (one for each possible letter/number we predict)
model.add(Dense(97, activation="softmax"))
# Ask Keras to build the TensorFlow model behind the scenes
model.compile(loss="categorical_crossentropy",
optimizer="adam",
metrics=["accuracy"])
# Train the neural network
model.fit(X_train,
Y_train,
validation_data=(X_test, Y_test),
batch_size=32,
epochs=num_epochs,
verbose=1)
# Save the trained model to disk
model.save(MODEL_FILENAME)
args = vars(ap.parse_args())
model = SeparableConv.build_model(width=48, height=48, depth=3, classes=2)
opt = Adagrad(lr=config.INIT_LR, decay=config.INIT_LR / config.NUM_EPOCHS)
model.compile(loss="binary_crossentropy", optimizer=opt, metrics=["accuracy"])
print(model.summary())
from keras.utils import plot_model
plot_model(model,
to_file='model_plot.png',
show_shapes=True,
show_layer_names=True)
# determine the total number of image paths in training, validation,
# and testing directories
trainPaths = list(paths.list_images(config.TRAIN_PATH))[:5000]
totalTrain = len(trainPaths)
totalVal = len(list(paths.list_images(config.VAL_PATH)))
totalTest = len(list(paths.list_images(config.TEST_PATH)))
print("totalTrain : % 2d, totalVal : % 2d, totalTest: % 2d" %
(totalTrain, totalVal, totalTest))
# account for skew in the labeled data
trainLabels = [int(p.split(os.path.sep)[-2]) for p in trainPaths]
trainLabels = np_utils.to_categorical(trainLabels)
classTotals = trainLabels.sum(axis=0)
classWeight = classTotals.max() / classTotals
# initialize the training training data augmentation object
trainAug = ImageDataGenerator(rescale=1 / 255.0,
# USAGE run in cmd: python3 encode_faces.py or python encode_faces.py
# import the necessary packages
from imutils import paths
import face_recognition
import pickle
import cv2
import os
# construct the argument parser and parse the arguments
# grab the paths to the input images in our dataset
print("[INFO] quantifying faces...")
imagePaths = list(paths.list_images("dataset"))
# initialize the list of known encodings and known names
knownEncodings = []
knownNames = []
# loop over the image paths
for (i, imagePath) in enumerate(imagePaths):
# extract the person name from the image path
print("[INFO] processing image {}/{}".format(i + 1,
len(imagePaths)))
name = imagePath.split(os.path.sep)[-2]
print(name)
# load the input image and convert it from RGB (OpenCV ordering)
# to dlib ordering (RGB)
image = cv2.imread(imagePath)
rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
def empty_download_folder():
imagePaths = list(paths.list_images(FTP_DIRECTORY))
for image in imagePaths:
os.remove(image)
def ListarImagenesDirectorio(dir):
return sorted(list(paths.list_images(dir + "/")))
# -*- coding: utf-8 -*-
"""
Created on Fri May 3 15:20:00 2018
@author: jercas
"""
from imutils import paths
import pandas as pd
import numpy as np
from icecream.icecream import ic
import shutil
import os
pathDirs = list(paths.list_images("./dataset/train_origin"))
data = np.loadtxt("train.txt", dtype=np.str_, delimiter=' ')
pathTxts = data[:, 0]
pathLabel = data[:, 1]
count = 1
for i, pathTxt in enumerate(pathTxts):
for j, pathDir in enumerate(pathDirs):
purePathDir = pathDir.split(os.path.sep)[-1]
if pathTxt == purePathDir:
ic(pathDir)
ic(pathTxt)
label = pathLabel[i]
if not os.path.exists("./dataset/train_split/{}".format(label)):
os.mkdir("./dataset/train_split/{}".format(label))
shutil.copyfile(
pathDir, "./dataset/train_split/{}/{}".format(label, pathTxt))
ic(count)
p = os.path.sep.join([args["output"], "{}.jpg".format(
str(total).zfill(8))])
f = open(p, "wb")
f.write(r.content)
f.close()
# update the counter
print("[INFO] downloaded: {}".format(p))
total += 1
# handle if any exceptions are thrown during the download process
except:
print("[INFO] error downloading {}...skipping".format(p))
# loop over the image paths we just downloaded
for imagePath in paths.list_images(args["output"]):
# initialize if the image should be deleted or not
delete = False
# try to load the image
try:
image = cv2.imread(imagePath)
# if the image is `None` then we could not properly load it
# from disk, so delete it
if image is None:
delete = True
# if OpenCV cannot load the image then the image is likely
# corrupt so we should delete it
except:
# initialize the HOG descriptor/person detector
hog = cv2.HOGDescriptor()
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector())
#video capture
vidcap = cv2.VideoCapture(r'C:\ADITYA\Computervision\UCF50_videos\UCF50\BaseballPitch\v_BaseballPitch_g01_c01.avi')
success,image = vidcap.read()
count = 0
while success:
cv2.imwrite(r"C:\ADITYA\Computervision\UCF50_videos\BaseballPitch_frames\frame%d.jpg" % count, image) # save frame as JPEG file
success,image = vidcap.read()
print('Read a new frame: ', success)
count += 1
# loop over the image paths
imagePaths = list(paths.list_images(args["images"]))
d=0
for imagePath in imagePaths:
# load the image and resize it to (1) reduce detection time
# and (2) improve detection accuracy
image = cv2.imread(imagePath)
image = imutils.resize(image, width=min(400, image.shape[1]))
orig = image.copy()
# detect people in the image
(rects, weights) = hog.detectMultiScale(image, winStride=(4, 4),
padding=(8, 8), scale=1.05)
# draw the original bounding boxes
for (x, y, w, h) in rects:
def prepare_cifar_data_for_ae2_svdd(train_path, test_path, modelpath):
NB_IV3_LAYERS_TO_FREEZE = 4
side = 32
side = 32
channel = 3
## Declare the scoring functions
g = lambda x: 1 / (1 + tf.exp(-x))
# g = lambda x : x # Linear
def nnScore(X, w, V, g):
# print "X",X.shape
# print "w",w[0].shape
# print "v",V[0].shape
return tf.matmul(g((tf.matmul(X, w))), V)
def relu(x):
y = x
y[y < 0] = 0
return y
def add_new_last_layer(base_model_output, base_model_input):
"""Add last layer to the convnet
Args:
base_model: keras model excluding top
nb_classes: # of classes
Returns:
new keras model with last layer
"""
x = base_model_output
print("base_model.output", x.shape)
inp = base_model_input
print("base_model.input", inp.shape)
dense1 = Dense(128, name="dense_output1")(x) # new sigmoid layer
dense1out = Activation("relu", name="output_activation1")(dense1)
dense2 = Dense(1, name="dense_output2")(dense1out) # new sigmoid layer
dense2out = Activation("relu", name="output_activation2")(
dense2) # new sigmoid layer
model = Model(inputs=inp, outputs=dense2out)
return model
# load the trained convolutional neural network freeze all the weights except for last four layers
print("[INFO] loading network...")
model = load_model(modelpath)
model = add_new_last_layer(model.output, model.input)
print(model.summary())
print("Length of model layers...", len(model.layers))
# Freeze the weights of untill the last four layers
for layer in model.layers[:NB_IV3_LAYERS_TO_FREEZE]:
layer.trainable = False
for layer in model.layers[NB_IV3_LAYERS_TO_FREEZE:]:
layer.trainable = True
print("[INFO] loading images for training...")
data = []
data_test = []
labels = []
labels_test = []
# grab the image paths and randomly shuffle them
imagePaths = sorted(list(paths.list_images(train_path)))
# loop over the input training images
image_dict = {}
test_image_dict = {}
i = 0
for imagePath in imagePaths:
# load the image, pre-process it, and store it in the data list
image = cv2.imread(imagePath)
image = cv2.resize(image, (side, side))
image = img_to_array(image)
data.append(image)
image_dict.update({i: imagePath})
i = i + 1
# extract the class label from the image path and update the
# labels list
label = imagePath.split(os.path.sep)[-2]
label = 1 if label == "dogs" else 0
labels.append(label)
# scale the raw pixel intensities to the range [0, 1]
data = np.array(data)
y = keras.utils.to_categorical(labels, num_classes=2)
# print image_dict
print("[INFO] preparing test data (anomalous )...")
testimagePaths = sorted(list(paths.list_images(test_path)))
# loop over the test images
j = 0
for imagePath in testimagePaths:
# load the image, pre-process it, and store it in the data list
image = cv2.imread(imagePath)
image = cv2.resize(image, (side, side))
image = img_to_array(image)
data_test.append(image)
test_image_dict.update({j: imagePath})
j = j + 1
# extract the class label from the image path and update the
# labels list
label = imagePath.split(os.path.sep)[-2]
label = 0 if label == "cats" else 1
labels_test.append(label)
# scale the raw pixel intensities to the range [0, 1]
data_test = np.array(data_test)
## Normalise the data
data = np.array(data) / 255.0
data_test = np.array(data_test) / 255.0
# Preprocess the inputs
x_train = data
x_test = data_test
## Obtain the intermediate output
layer_name = 'dense_1'
intermediate_layer_model = Model(
inputs=model.input, outputs=model.get_layer(layer_name).output)
layer1 = model.get_layer("dense_output1")
layer2 = model.get_layer("dense_output2")
intermediate_output = intermediate_layer_model.predict(x_train)
X = intermediate_output
intermediate_output = intermediate_layer_model.predict(x_test)
X_test = intermediate_output
K.clear_session()
return [X, X_test]
hist = hist.astype("float")
hist /= (hist.sum() + eps)
# return the histogram of Local Binary Patterns
return hist
#%%
# initialize the local binary patterns descriptor along with
# the data and label lists
desc = LocalBinaryPatterns(24, 8)
data = []
labels = []
#%%
# loop over the training images
for imagePath in paths.list_images('/home/alfredo/Desktop/PROJECTS/Master Thesis/opencv/images/local_binary_patterns/training'):
# load the image, convert it to grayscale, and describe it
image = cv2.imread(imagePath)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
hist = desc.describe(gray)
# extract the label from the image path, then update the
# label and data lists
labels.append(imagePath.split(os.path.sep)[-2])
data.append(hist)
#%%
# train a Linear SVM on the data
model = LinearSVC(C=100.0, random_state=42)
model.fit(data, labels)
ap.add_argument("-p",
"--plot",
type=str,
default="plot.png",
help="path to output loss/accuracy plot")
args = vars(ap.parse_args())
EPOCHS = 25
INIT_LR = 1e-3
BS = 32
print("[INFO] loading images...")
data = []
labels = []
imagePaths = sorted(list(paths.list_images(args["dataset"])))
random.seed(42)
random.shuffle(imagePaths)
for imagePath in imagePaths:
image = cv2.imread(imagePath)
image = cv2.resize(image, (28, 28))
image = img_to_array(image)
data.append(image)
label = imagePath.split(os.path.sep)[-2]
label = 1 if label == "fire" else 0
labels.append(label)
data = np.array(data, dtype="float") / 255.0
labels = np.array(labels)
import imutils
import cv2
import random
import pickle
# Construct the argument parser and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-m", "--model", required=True, help="path to trained model")
ap.add_argument("-t", "--testset", required=True, help="path to test images")
args = vars(ap.parse_args())
# Resize parmeters (RESIZE should be the same as used in training)
RESIZE = 96
# Grab the image paths and randomly shuffle them
image_paths = sorted(list(paths.list_images(args["testset"])))
random.seed()
random.shuffle(image_paths)
# Read labels for classes to recognize
mlb = pickle.loads(open(args["model"] + ".lbl", "rb").read())
labels = mlb.classes_
# Load the trained network
model = load_model(args["model"] + ".h5")
print(model.summary())
for image_path in image_paths:
# Load the image
image = cv2.imread(image_path)
output = imutils.resize(image, width=400)
from imutils import paths
# Config
crop_start = (50, 285)
crop_size = (224, 224)
dark_factor = 0.3
# Prepare crop area
crop_area = numpy.index_exp[
crop_start[0]:crop_start[0]+crop_size[0],
crop_start[1]:crop_start[1]+crop_size[1]
]
# find the image paths and randomly shuffle them
print("# loading images...")
imagePaths = sorted(list(paths.list_images(sys.argv[1], contains="capture")))
print("# loaded %d images." % (len(imagePaths)))
for imagePath in imagePaths:
image = cv2.imread(imagePath)
orig = image[crop_area]
image = (image * dark_factor).astype(numpy.uint8)
image[crop_area] = orig
cv2.imshow("Output", image)
key_raw = cv2.waitKey(0)
try:
key = chr(key_raw)
if key in ['d', 'D']:
print("delete:%s" % imagePath)
os.unlink(imagePath)
elif key in ['p', 'P', 'y', 'Y']:
from keras.layers.core import Flatten, Dense
from helpers import resize_to_fit
from keras.callbacks import ModelCheckpoint, TensorBoard
from keras.utils.vis_utils import plot_model
# LETTER_IMAGES_FOLDER = "extracted_letter_images"
LETTER_IMAGES_FOLDER = "extracted_letter_images"
MODEL_FILENAME = "captcha_model.hdf5"
MODEL_LABELS_FILENAME = "model_labels.dat"
# initialize the data and labels
data = []
labels = []
# loop over the input images
for image_file in paths.list_images(LETTER_IMAGES_FOLDER):
# Load the image and convert it to grayscale
image = cv2.imread(image_file)
try:
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
except Exception:
print(image)
# Resize the letter so it fits in a 20x20 pixel box
image = resize_to_fit(image, 20, 20)
# Add a third channel dimension to the image to make Keras happy
image = np.expand_dims(image, axis=2)
# Grab the name of the letter based on the folder it was in
label = image_file.split(os.path.sep)[-2]
def train(number_neighbors, focus, hidden_layer_sizes, max_iter_bpnn,
max_iter_svm):
respond = {}
__init_log(focus)
respond['focus'] = focus
imagePaths = list(paths.list_images(FOLDER_DATASET))
rawImages = []
features = []
labels = []
for (i, imagePath) in enumerate(imagePaths):
image = cv2.imread(imagePath)
label = imagePath.split(os.path.sep)[-1].split(".")[0]
pixels = __image_to_feature_vector(image)
hist = __extract_color_histogram(image)
rawImages.append(pixels)
features.append(hist)
labels.append(label)
if i > 0 and ((i + 1) % 200 == 0 or i == len(imagePaths) - 1):
size_images = len(imagePaths)
line = "[INFO] processed {}/{}".format(i + 1, size_images)
respond['length_images'] = size_images
__write_log(line)
labels = np.array(labels)
if focus == 'histogram':
features = np.array(features)
size_package = '{:.2f}MB'.format(features.nbytes / (1024 * 1000.0))
line = '[INFO] features matrix: ' + size_package
respond['size_package'] = size_package
__write_log(line)
(trainFeat, testFeat, trainLabels,
testLabels) = train_test_split(features,
labels,
test_size=0.15,
random_state=42)
# k-NN
line = "[INFO] evaluating histogram accuracy..."
__write_log(line)
model = KNeighborsClassifier(n_neighbors=number_neighbors)
model.fit(trainFeat, trainLabels)
acc = model.score(testFeat, testLabels)
line = "[INFO] k-NN classifier: k=%d" % number_neighbors
__write_log(line)
accuracy_knn = '{:.2f}%'.format(acc * 100)
line = "[INFO] histogram accuracy: " + accuracy_knn
respond['accuracy_knn'] = accuracy_knn
__write_log(line)
joblib.dump(model, FILE_MODELS_HISTOGRAM_KNN)
# neural network
line = "[INFO] evaluating histogram accuracy..."
__write_log(line)
model = MLPClassifier(hidden_layer_sizes=(50, ),
max_iter=1000,
alpha=1e-4,
solver='sgd',
tol=1e-4,
random_state=1,
learning_rate_init=.1)
model.fit(trainFeat, trainLabels)
acc = model.score(testFeat, testLabels)
accuracy_bpnn = "{:.2f}%".format(acc * 100)
line = "[INFO] neural network histogram accuracy: " + accuracy_bpnn
respond['accuracy_bpnn'] = accuracy_bpnn
__write_log(line)
joblib.dump(model, FILE_MODELS_HISTOGRAM_BPNN)
# SVC
line = "[INFO] evaluating histogram accuracy..."
__write_log(line)
model = SVC(max_iter=1000, class_weight='balanced')
model.fit(trainFeat, trainLabels)
acc = model.score(testFeat, testLabels)
accuracy_svm = '{:.2f}%'.format(acc * 100)
line = "[INFO] SVM-SVC histogram accuracy: " + accuracy_svm
respond['accuracy_svm'] = accuracy_svm
__write_log(line)
joblib.dump(model, FILE_MODELS_HISTOGRAM_SVM)
respond['success'] = True
return respond
elif focus == 'pixel':
rawImages = np.array(rawImages)
size_package = "{:.2f}MB".format(rawImages.nbytes / (1024 * 1000.0))
line = "[INFO] pixels matrix: " + size_package
respond['size_package'] = size_package
__write_log(line)
(trainRI, testRI, trainRL, testRL) = train_test_split(rawImages,
labels,
test_size=0.15,
random_state=42)
# k-NN
line = "[INFO] evaluating raw pixel accuracy..."
__write_log(line)
model = KNeighborsClassifier(n_neighbors=number_neighbors)
model.fit(trainRI, trainRL)
acc = model.score(testRI, testRL)
line = "[INFO] k-NN classifier: k=%d" % number_neighbors
__write_log(line)
accuracy_knn = "{:.2f}%".format(acc * 100)
line = "[INFO] raw pixel accuracy: " + accuracy_knn
respond['accuracy_knn'] = accuracy_knn
__write_log(line)
joblib.dump(model, FILE_MODELS_PIXEL_KNN)
# neural network
line = "[INFO] evaluating raw pixel accuracy..."
__write_log(line)
model = MLPClassifier(hidden_layer_sizes=(hidden_layer_sizes, ),
max_iter=max_iter_bpnn,
alpha=1e-4,
solver='sgd',
tol=1e-4,
random_state=1,
learning_rate_init=.1)
model.fit(trainRI, trainRL)
acc = model.score(testRI, testRL)
accuracy_bpnn = "{:.2f}%".format(acc * 100)
line = "[INFO] neural network raw pixel accuracy: " + accuracy_bpnn
respond['accuracy_bpnn'] = accuracy_bpnn
__write_log(line)
joblib.dump(model, FILE_MODELS_PIXEL_BPNN)
# SVC
line = "[INFO] evaluating raw pixel accuracy..."
__write_log(line)
model = SVC(max_iter=max_iter_svm, class_weight='balanced')
model.fit(trainRI, trainRL)
acc = model.score(testRI, testRL)
accuracy_svm = "{:.2f}%".format(acc * 100)
line = "[INFO] SVM-SVC raw pixel accuracy: " + accuracy_svm
respond['accuracy_svm'] = accuracy_svm
__write_log(line)
joblib.dump(model, FILE_MODELS_PIXEL_SVM)
respond['success'] = True
return respond
else:
respond['success'] = False
respond['error'] = 'focus no esperado'
return respond
# -*- coding: utf-8 -*-
"""
Created on Thu Jul 9 04:16:07 2020
@author: Devdarshan
"""
import os
from imutils import paths
path = "Fire/"
path2 = "images/"
imagePaths = list(paths.list_images(path))
import cv2
i = 0
for imagePath in imagePaths:
# load the image and resize it to be a fixed 128x128 pixels,
# ignoring aspect ratio
print(imagePath)
image = cv2.imread(imagePath)
image = cv2.resize(image, (608, 608))
cv2.imwrite(os.path.join(path2, "{}.jpg").format(i), image)
i += 1
f_ishift = numpy.fft.ifftshift(fshift)
img_fft = numpy.fft.ifft2(f_ishift)
img_fft = 20*numpy.log(numpy.abs(img_fft))
result = numpy.mean(img_fft)
return result
# construct the argument parse and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-i", "--images", required=True,
help="path to input directory of images")
args = vars(ap.parse_args())
# loop over the input images
for i, imagePath in enumerate(sorted(paths.list_images(args["images"]))):
# load the image, convert it to grayscale, and compute the
# focus measure of the image using the Variance of Laplacian
# method
image = cv2.imread(imagePath)
image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
eqhist = cv2.equalizeHist(gray)
metrics = {}
width, height, _ = image.shape
cropped = image[50:height-50,50:width-50,:]
grayc = cv2.cvtColor(cropped, cv2.COLOR_RGB2GRAY)
# import the necessary packages
from imutils import paths
import argparse
import dlib
import cv2
# construct the argument parse and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-d", "--detector", required=True, help="Path to trained object detector")
ap.add_argument("-t", "--testing", required=True, help="Path to directory of testing images")
args = vars(ap.parse_args())
# load the detector
detector = dlib.simple_object_detector(args["detector"])
# loop over the testing images
for testingPath in paths.list_images(args["testing"]):
# load the image and make predictions
image = cv2.imread(testingPath)
boxes = detector(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
# loop over the bounding boxes and draw them
for b in boxes:
(x, y, w, h) = (b.left(), b.top(), b.right(), b.bottom())
cv2.rectangle(image, (x, y), (w, h), (0, 255, 0), 2)
print (x,y)
# show the image
cv2.imshow("Image", image)
cv2.waitKey(0)
import cv2
import numpy as np
from imutils import paths
import os
from sklearn.metrics import accuracy_score
from matplotlib import pyplot as plt
classes = {1: 'Bike',
2: 'Horse'}
labels = []
dictionarySize = 12
BOW = cv2.BOWKMeansTrainer(dictionarySize)
sift = cv2.xfeatures2d.SIFT_create()
imagePaths = list(paths.list_images("train"))
for image in imagePaths:
label = image.split('/')[1]
if label == 'Bike':
labels.append(1)
elif label == 'Horse':
labels.append(2)
image = image.replace("\\","")
img = cv2.imread(image,0)
kp, des = sift.detectAndCompute(img,None)
BOW.add(des)
dictionary = BOW.cluster()
print("Created Bag of Words")
ap.add_argument("-l", "--labelbin", required=True,
help="path to output label binarizer")
ap.add_argument("-p", "--plot", type=str, default="plot.png",
help="path to output accuracy/loss plot")
args = vars(ap.parse_args())
# initialize the number of epochs to train for, initial learning rate,
# batch size, and image dimensions
EPOCHS = 75
INIT_LR = 1e-3
BS = 32
IMAGE_DIMS = (96, 96, 3)
# grab the image paths and randomly shuffle them
print("[INFO] loading images...")
imagePaths = sorted(list(paths.list_images(args["dataset"])))
random.seed(42)
random.shuffle(imagePaths)
# initialize the data and labels
data = []
labels = []
# loop over the input images
for imagePath in imagePaths:
# load the image, pre-process it, and store it in the data list
image = cv2.imread(imagePath)
image = cv2.resize(image, (IMAGE_DIMS[1], IMAGE_DIMS[0]))
image = img_to_array(image)
data.append(image)
help="Path to the CALTECH-101 class annotations")
ap.add_argument("-o",
"--output",
required=True,
help="Path to the output detector")
args = vars(ap.parse_args())
# grab the default training options for our HOG + Linear SVM detector initialize the
# list of images and bounding boxes used to train the classifier
print("[INFO] gathering images and bounding boxes...")
options = dlib.simple_object_detector_training_options()
images = []
boxes = []
# loop over the image paths
for imagePath in paths.list_images(args["class"]):
# extract the image ID from the image path and load the annotations file
imageID = imagePath[imagePath.rfind("/") + 1:].split("_")[1]
imageID = imageID.replace(".jpg", "")
p = "{}/annotation_{}.mat".format(args["annotations"], imageID)
annotations = loadmat(p)["box_coord"]
# loop over the annotations and add each annotation to the list of bounding
# boxes
bb = [
dlib.rectangle(left=long(x),
top=long(y),
right=long(w),
bottom=long(h)) for (y, h, x, w) in annotations
]
boxes.append(bb)
import numpy as np
import argparse
import imutils
import cv2
# construct the argument parse and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-i", "--images", required=True, help="path to images directory")
args = vars(ap.parse_args())
# initialize the HOG descriptor/person detector
hog = cv2.HOGDescriptor()
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector())
# loop over the image paths
for imagePath in paths.list_images(args["images"]):
# load the image and resize it to (1) reduce detection time
# and (2) improve detection accuracy
image = cv2.imread(imagePath)
image = imutils.resize(image, width=min(400, image.shape[1]))
orig = image.copy()
# detect people in the image
(rects, weights) = hog.detectMultiScale(image, winStride=(4, 4),
padding=(8, 8), scale=1.05)
# draw the original bounding boxes
for (x, y, w, h) in rects:
cv2.rectangle(orig, (x, y), (x + w, y + h), (0, 0, 255), 2)
# apply non-maxima suppression to the bounding boxes using a
ap = argparse.ArgumentParser()
ap.add_argument("-b", "--base-model", required=True,
help="base model path")
ap.add_argument("-i", "--images", required=True,
help="base path to input directory of images")
ap.add_argument("-o", "--output", required=True,
help="base path to output directory")
ap.add_argument("-l", "--layers", nargs='+', default=["conv2/3x3", "inception_3b/5x5_reduce", "inception_4c/output"],
help="layer or layers to use")
args = ap.parse_args()
# buy the ticket, take the ride
bc = BatCountry(args.base_model)
# loop over the input directory of images
for imagePath in paths.list_images(args.images):
# loop over the layers
for layer in args.layers:
# we can't stop here...
print("[INFO] processing `{}`".format(imagePath))
image = bc.dream(np.float32(Image.open(imagePath)), end=layer)
# write the output image to file
filename = imagePath[imagePath.rfind("/") + 1:]
outputPath = "{}/{}_{}".format(args.output, layer.replace("/", "_"), filename)
result = Image.fromarray(np.uint8(image))
result.save(outputPath)
# do some cleanup
bc.cleanup()
import cv2
ap = argparse.ArgumentParser()
ap.add_argument("-d", "--dataset", required=True)
ap.add_argument("-k", "--clusters", type=int, default=2,
help="# of clusters to generate")
args = vars(ap.parse_args())
# initialize the image descriptor along with the image matrix
desc = LabHistogram([8, 8, 8])
data = []
# grab image paths from dataset directory
image_paths = list(paths.list_images(args["dataset"]))
image_paths = np.array(sorted(image_paths))
# loop over the input dataset of images
for im_path in image_paths:
# load the image, describe the image, then update the list of data
image = cv2.imread(im_path)
hist = desc.describe(image)
data.append(hist)
# cluster the color histograms
clt = KMeans(n_clusters=args["clusters"])
labels = clt.fit_predict(data)
from IPython import embed
def move(path):
for pf in list_images(path):
f = pf.split('/')[-1]
new_f = f[0] + ("%05d" % int(f[1:-4])) + f[-4:]
shutil.move(pf, os.path.join(path, new_f))
# Freeze the weights of untill the last four layers
for layer in model.layers[:NB_IV3_LAYERS_TO_FREEZE]:
layer.trainable = False
for layer in model.layers[NB_IV3_LAYERS_TO_FREEZE:]:
layer.trainable = True
# declare the global variable
print("[INFO] loading images for training...")
data = []
data_test = []
labels = []
labels_test = []
# grab the image paths and randomly shuffle them
imagePaths = sorted(list(paths.list_images(train_path)))
import random
random.seed(42)
random.shuffle(imagePaths)
# loop over the input training images
image_dict = {}
i = 0
for imagePath in imagePaths:
# load the image, pre-process it, and store it in the data list
image = cv2.imread(imagePath)
image = cv2.resize(image, (28, 28))
image = img_to_array(image)
data.append(image)
image_dict.update({i: imagePath})
i = i + 1
img_rows, img_cols = 28, 28
# number of convolutional filters to use
nb_filters = 32
# size of pooling area for max pooling
pool_size = (2, 2)
# convolution kernel size
kernel_size = (3, 3)
# the data, shuffled and split between train and test sets
# (X_train, y_train), (X_test, y_test) = mnist.load_data()
X_train = np.empty((0, 1, img_rows, img_cols))
y_train = np.empty((0))
X_test = np.empty((0, 1, img_rows, img_cols))
y_test = np.empty((0))
for i in range(0, 10):
(X, Y) = build_batch(sorted(list(paths.list_images("dataset/{}".format(i)))), img_rows, img_cols)
X_train = np.concatenate((np.asarray(X[0:75]), X_train))
y_train = np.concatenate((np.asarray(Y[0:75]), y_train))
X_test = np.concatenate((np.asarray(X[75:]), X_test))
y_test = np.concatenate((np.asarray(Y[75:]), y_test))
print(X_train.shape)
print(y_train.shape)
print(X_test.shape)
print(y_test.shape)
if K.image_dim_ordering() == 'th':
X_train = X_train.reshape(X_train.shape[0], 1, img_rows, img_cols)
X_test = X_test.reshape(X_test.shape[0], 1, img_rows, img_cols)
input_shape = (1, img_rows, img_cols)
else:
from localbinarypatterns import LocalBinaryPatterns
from imutils import paths
import cv2
import pandas as pd
import numpy as np
from skimage import feature
import matplotlib.pyplot as plt
db = 'yale'
data_training = 'db/%s' % db
P = 8 # número de pontos
R = 8 # raio
dataset = [file for file in paths.list_images(data_training)]
desc = LocalBinaryPatterns(8, 8)
gbr = paths.list_images(data_training)
nama_gbr = gbr.__next__()
print(nama_gbr)
image = cv2.imread(nama_gbr)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
lbp = feature.local_binary_pattern(gray, P, R, method='uniform')
hist = np.histogram(lbp.ravel(), bins=range(0, P + 3))
h = hist[0].astype('float')
h /= (h.sum())
print(h, h.sum())
plt.bar(np.array(range(0, P + 2)), h)
plt.xticks(np.array(range(0, P + 2)))
plt.ylabel('Percentage')
plt.show()
if self.preprocessors is not None:
for p in self.preprocessors:
image = p.preprocess(image)
#print(image.shape)
data.append(image)
labels.append(label)
if verbose > 0 and i>0 and (i+1)%verbose==0:
print("[INFO] processed {}/{}".format(i+1, len(imagePaths)))
return(np.array(data), np.array(labels))
# In[57]:
image_paths = list(paths.list_images('/home/padmach/data/DogVsCats/train'))
# In[58]:
sp = SimplePreprocessor(32, 32)
sdl = SimpleDatasetLoader(preprocessors=[sp])
(data, labels) = sdl.load(image_paths, verbose=500)
# In[60]:
data = data.reshape((data.shape[0], 3072))
# In[63]:
from imutils import paths
import numpy as np
import imutils
import cv2
crop = 1
print("[INFO] loading images...")
imagePaths = sorted(list(paths.list_images('images')))
print(imagePaths)
images = []
for imagePath in imagePaths:
image = cv2.imread(imagePath)
images.append(image)
print("[INFO] stitching images...")
stitcher = cv2.createStitcher() if imutils.is_cv3() else cv2.Stitcher_create()
(status, stitched) = stitcher.stitch(images)
if status == 0:
if crop > 0:
print("[INFO] cropping...")
stitched = cv2.copyMakeBorder(stitched, 10, 10, 10, 10,
cv2.BORDER_CONSTANT, (0, 0, 0))
gray = cv2.cvtColor(stitched, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY)[1]
f = open(p, "wb")
f.write(r.content)
f.close()
# update the counter
print("[INFO] downloaded: {}".format(p))
total += 1
# handle if any exceptions are thrown during the download process
except:
print("[INFO] error downloading {}...skipping".format(p))
# loop over the image paths we just downloaded
for imagePath in paths.list_images(args["output"]):
# initialize if the image should be deleted or not
delete = False
# try to load the image
try:
image = cv2.imread(imagePath)
# if the image is `None` then we could not properly load it
# from disk, so delete it
if image is None:
delete = True
# if OpenCV cannot load the image then the image is likely
# corrupt so we should delete it
except:
cv2.namedWindow("quadro")
cv2.setMouseCallback("quadro",alteraEstado,param=(lista,img))
cv2.imshow("quadro",img)
print conta,'Arq:',lista[conta]["filename"],' Sel Status:',lista[conta]["isok"]
if __name__ == '__main__':
cascade = cv2.CascadeClassifier(args.cascade)
largeimg = np.zeros((800,1280,3),np.uint8)
xpos = 0
ypos = 0
listaInfoJ = []
conta = 0
for pimagem in paths.list_images(mypath):
print pimagem
rgbIn = cv2.imread(pimagem)
[heb,wib,pb] = rgbIn.shape
rects,bdisp=detect(rgbIn, cascade)
for x1, y1, x2, y2 in rects:
x1,y1,x2,y2 = RepUtil.novoEquad(x1,y1,x2,y2,wib,heb)
vis_roi = rgbIn[y1:y2, x1:x2]
[he,wi,pro] = vis_roi.shape
if wi > 640:
fato = 640.0/float(wi)
rgbFrame = cv2.resize(vis_roi,(0,0),fx=fato,fy=fato)
else:
rgbFrame = vis_roi.copy()
bb = align.getLargestFaceBoundingBox(rgbFrame)
bbs = [bb] if bb is not None else []
model.compile(optimizer='adam',
loss='categorical_crossentropy',
metrics=['accuracy'])
print("[INFO] Model Compiled!")
return model
#%%
print("[INFO] loading images from private data...")
data = []
labels = []
# grab the image paths and randomly shuffle them
imagePaths = sorted(
list(paths.list_images('C:\\Users\\User\\spn_myone\\spn_pr'))
) # data folder with 2 categorical folders
random.seed(SEED)
random.shuffle(imagePaths)
# loop over the input images
for imagePath in imagePaths:
# load the image, resize the image to be 32x32 pixels (ignoring aspect ratio),
# flatten the 32x32x3=3072 pixel image into a list, and store the image in the data list
image = cv2.imread(imagePath)
#image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
image = cv2.resize(image, (32, 32)) / 255
data.append(image)
# extract the class label from the image path and update the labels list
label = imagePath.split(os.path.sep)[-2]
print "{}".format(rect)
x1,y1,x2,y2 = novoEquadR(rect,w,h)
print "{}; {}; {}; {}".format(x1,y1,x2,y2)
newim = img[y1:y2,x1:x2].copy()
bb = align.getLargestFaceBoundingBox(newim)
if bb is None:
continue
hr,wr,c = newim.shape
if wr > 600:
newim = cv2.resize(newim,(0,0),fx=0.5,fy=0.5)
return newim
return None
if __name__ == '__main__':
cascade = cv2.CascadeClassifier(join("..","..","data","haarcascades","haarcascade_frontalface_alt2.xml"))
for pimagem in paths.list_images(pathimg):
print pimagem
im = cv2.imread(pimagem)
if im is None:
continue
im=detect(im, cascade)
if im is None:
continue
h,w,c = im.shape
if h < 100 or w < 100:
print "Imagem pequena??? -> {} -> {}x{}".format(pimagem,w,h)
continue
if im is None:
import numpy as np
import argparse
import imutils
import cv2
# construct the argument parse and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-i", "--images", required=True, help="path to images directory")
args = vars(ap.parse_args())
# initialize the HOG descriptor/person detector
hog = cv2.HOGDescriptor()
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector())
# loop over the image paths
imagePaths = list(paths.list_images(args["images"]))
for imagePath in imagePaths:
# load the image and resize it to (1) reduce detection time
# and (2) improve detection accuracy
image = cv2.imread(imagePath)
image = imutils.resize(image, width=min(400, image.shape[1]))
orig = image.copy()
# detect people in the image
(rects, weights) = hog.detectMultiScale(image, winStride=(4, 4),
padding=(8, 8), scale=1.05)
# draw the original bounding boxes
for (x, y, w, h) in rects:
cv2.rectangle(orig, (x, y), (x + w, y + h), (0, 0, 255), 2)
# author: Adrian Rosebrock
# website: http://www.pyimagesearch.com
# import the necessary packages
from __future__ import print_function
from imutils import paths
# loop over the image paths in the previous 'demo_images'
# directory and print the paths to the terminal
for imagePath in paths.list_images("../demo_images"):
print(imagePath)
MODEL_FILENAME = "captcha_model.hdf5"
MODEL_LABELS_FILENAME = "model_labels.dat"
CAPTCHA_IMAGE_FOLDER = "generated_captcha_images"
# Load up the model labels (so we can translate model predictions to actual letters)
with open(MODEL_LABELS_FILENAME, "rb") as f:
lb = pickle.load(f)
# Load the trained neural network
model = load_model(MODEL_FILENAME)
# Grab some random CAPTCHA images to test against.
# In the real world, you'd replace this section with code to grab a real
# CAPTCHA image from a live website.
captcha_image_files = list(paths.list_images(CAPTCHA_IMAGE_FOLDER))
captcha_image_files = np.random.choice(captcha_image_files, size=(100,), replace=False)
# loop over the image paths
for image_file in captcha_image_files:
# Load the image and convert it to grayscale
image = cv2.imread(image_file)
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
# Add some extra padding around the image
image = cv2.copyMakeBorder(image, 20, 20, 20, 20, cv2.BORDER_REPLICATE)
# threshold the image (convert it to pure black and white)
thresh = cv2.threshold(image, 0, 255, cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)[1]
# find the contours (continuous blobs of pixels) the image
def readImagesAndAnnotations(self):
self.imagePaths = list(paths.list_images(self.inputPath))
import config
from imutils import paths
import random
import shutil
import os
# Grab the paths to all input images in the original input directory and shuffle them
imagePaths = list(paths.list_images(config.ORIG_INPUT_DATASET))
random.seed(100)
random.shuffle(imagePaths)
# Compute the training and testing split
i = int(len(imagePaths) * config.TRAIN_SPLIT)
trainPaths = imagePaths[:i]
testPaths = imagePaths[i:]
# Using part of the training dataa for validation
val_num = int(len(trainPaths) * config.VAL_SPLIT)
valPaths = trainPaths[:i]
trainPaths = trainPaths[i:]
# Define the datasets that we'll be building
datasets = [("training", trainPaths, config.TRAIN_PATH),
("validation", valPaths, config.VALIDATION_PATH),
("testing", testPaths, config.TEST_PATH)]
# Loop over the datasets
for (dataType, imagePaths, baseOutput) in datasets:
# Show which dataset we are creating
print("[INFO] Buildint {} split.".format(dataType))
def updata(identity,path,base,oldFname,separator="\\"):
listIm = paths.list_images(path)
contai = 0
for fileim in listIm:
vf = fileim.split(separator)
#print vf[len(vf)-1]
if oldFname is not None:
if fileim in oldFname:
print "O arquivo {} tem duplicacao".format(fileim)
continue
#print "Processa {}".format(fileim)
img = cv2.imread(fileim)
mdb.gravaIm(img, base, contai, identity,vf[len(vf)-1], ".png")
contai += 1
return contai