def generator(samples, batch_size=32):
num_samples = len(samples)
while 1:
shuffle(samples)
for offset in range(0, num_samples, batch_size):
batch_samples = samples[offset:offset + batch_size]
# for each batch, open the image file to RGB, remove any noise
# with gaussian blur and resize the image to fit into neural network
images = []
angles = []
for batch_sample in batch_samples:
center_image = read_image_rgb(batch_sample[0])
center_image = remove_noise(center_image)
center_image = resize_image(center_image)
measurement = float(batch_sample[1])
images.append(center_image)
angles.append(measurement)
flipped_image, flipped_measurement = flip_image(
center_image, measurement)
images.append(flipped_image)
angles.append(flipped_measurement)
X_train = np.array(images)
y_train = np.array(angles)
yield shuffle(X_train, y_train)
def POST(self, name = None):
xinput = web.input()
ifile = web.input(avatar={})
filePath_a = goods_image_name()
filePath = filePath_a[0]
if 'avatar' in ifile:
ifile['avatar'].file.seek(0, os.SEEK_END)
if ifile.avatar.file.tell() > 0:
fout = open(filePath % '_o','w') # creates the file where the uploaded file should be stored
ifile.avatar.file.seek(0)
fout.write(ifile.avatar.file.read()) # writes the uploaded file to the newly created file.
fout.close() # closes the file, upload complete.
resize_image(filePath % '_o', filePath % '_256', 256, refMode = 'both')
resize_image(filePath % '_o', filePath % '_512', 512, refMode = 'both')
resize_image(filePath % '_o', filePath % '_800', 800, refMode = 'both')
thegoods = goods()
thegoods.bind(xinput)
thegoods.avatar_path.zval = filePath_a[1]
thegoods.update_date.zval = datetime.datetime.now()
#return thegoods.dump()
if name is None:
thegoods.create_date.zval = datetime.datetime.now()
insert(thegoods)
else:
thegoods._id.val(name)
update(thegoods)
raise web.seeother("/admin/goods/")
def recognize(self,
images,
detection_kwargs=None,
recognition_kwargs=None):
"""Run the pipeline on one or multiples images.
Args:
images: The images to parse (can be a list of actual images or a list of filepaths)
detection_kwargs: Arguments to pass to the detector call
recognition_kwargs: Arguments to pass to the recognizer call
Returns:
A list of lists of (text, box) tuples.
"""
# Make sure we have an image array to start with.
if not isinstance(images, np.ndarray):
images = [tools.read(image) for image in images]
# This turns images into (image, scale) tuples temporarily
images = [
tools.resize_image(image,
max_scale=self.scale,
max_size=self.max_size) for image in images
]
max_height, max_width = np.array(
[image.shape[:2] for image, scale in images]).max(axis=0)
scales = [scale for _, scale in images]
images = np.array([
tools.pad(image, width=max_width, height=max_height)
for image, _ in images
])
if detection_kwargs is None:
detection_kwargs = {}
if recognition_kwargs is None:
recognition_kwargs = {}
box_groups = self.detector.detect(images=images, **detection_kwargs)
prediction_groups = self.recognizer.recognize_from_boxes(
images=images, box_groups=box_groups, **recognition_kwargs)
box_groups = [
tools.adjust_boxes(
boxes=boxes, boxes_format='boxes', scale=1 /
scale) if scale != 1 else boxes
for boxes, scale in zip(box_groups, scales)
]
return [
list(zip(predictions, boxes))
for predictions, boxes in zip(prediction_groups, box_groups)
]
def POST(self, name = None):
xinput = web.input()
if xinput.has_key('act') and xinput.act == 'del':
imgid = xinput.id
mgoodsImg = find_one(goods_picture, _id=imgid)
xpath = mgoodsImg.path.zval
try:
del_goods_images(xpath)
except OSError:
pass
remove(goods_picture, _id=imgid)
return 1
elif xinput.has_key('act') and xinput.act == 'avatar':
imgid = xinput.id
mgoodsImg = find_one(goods_picture, _id=imgid)
xpath = mgoodsImg.path.zval
xgoods_id = mgoodsImg.goods_id.zval
mGoods = goods()
mGoods._id.zval = xgoods_id
mGoods.avatar_path.zval = xpath
update(mGoods)
return 1
ifile = web.input(avatar={})
filePath_a = goods_image_name()
filePath = filePath_a[0]
if 'avatar' in ifile:
ifile.avatar.file.seek(0, os.SEEK_END)
if ifile.avatar.file.tell() > 0:
fout = open(filePath % '_o','w') # creates the file where the uploaded file should be stored
ifile.avatar.file.seek(0)
fout.write(ifile.avatar.file.read()) # writes the uploaded file to the newly created file.
fout.close() # closes the file, upload complete.
resize_image(filePath % '_o', filePath % '_256', 256, refMode = 'both')
resize_image(filePath % '_o', filePath % '_512', 512, refMode = 'both')
resize_image(filePath % '_o', filePath % '_800', 800, refMode = 'both')
thepic = goods_picture()
thepic.bind(xinput)
thepic.path.zval = filePath_a[1]
thepic.goods_id.zval = name
thepic.create_date.zval = datetime.datetime.now()
insert(thepic)
raise web.seeother("/admin/goods/images/" + name)
if __name__ == '__main__':
tf.config.list_physical_devices('GPU')
args, _ = parse_args()
FLAGS['process_max_epoch'] = args.epochs
FLAGS['folder_model']= args.model_dir
FLAGS['gpu']= args.n_gpu
FLAGS['input_folder']= args.train_input
FLAGS['label_folder']= args.label_input
tools.resize_image(FLAGS['input_folder'],max_length = 512)
tools.resize_image(FLAGS['label_folder'],max_length = 512)
tf.compat.v1.disable_eager_execution()
print(tf.config.list_physical_devices('GPU'))
#tf.compat.v1.disable_resource_variables()
os.environ["CUDA_VISIBLE_DEVICES"] = FLAGS['num_gpu']
sys.stdout = Tee(sys.stdout, open(FLAGS['txt_log'], 'a+'))
LOG_STAMP_TRAIN = np.int32(np.linspace(0, FLAGS['data_train_batch_count']-1, num=FLAGS['process_train_log_interval_epoch']+1))
LOG_STAMP_TEST = np.int32(np.linspace(0, FLAGS['data_train_batch_count']-1, num=FLAGS['process_test_log_interval_epoch'] +1))
def netG_concat_value(tensor, v):
v_t = tf.constant(v, dtype=tf.float32, shape=tensor.get_shape().as_list()[:3] + [1])
tensor = tf.concat(3, [tensor, v_t])
return tensor
input_img = input_img[None, :, :, :]
dict_d = [input_img, rect, 0]
dict_t = [test_df.input1_src] + \
test_df.mat1.rect + test_df.mat1.rot
enhance_test_img = sess.run(
netG_test_output1_crop,
feed_dict={t: d
for t, d in zip(dict_t, dict_d)})
enhance_test_img = safe_casting(
enhance_test_img * tf.as_dtype(FLAGS['data_input_dtype']).max,
FLAGS['data_input_dtype'])
enhanced_img_file_name = file_out_name_without_ext + FLAGS[
'data_output_ext']
enhance_img_file_path = FLAGS['folder_test_img'] + enhanced_img_file_name
#try:
# print(current_time() + ', try remove file path = %s' % enhance_img_file_path)
# os.remove(enhance_img_file_path)
#except OSError as e:
# print(current_time() + ', remove fail, error = %s' % e.strerror)
cv2.imwrite(enhance_img_file_path, enhance_test_img)
return enhanced_img_file_name
if __name__ == '__main__':
args, _ = parse_args()
FLAGS['inference_folder'] = args.inference_dir
tools.resize_image(FLAGS['inference_folder'])
print('processing')
processImg('a0002.tif', 'totaltest')
if __name__ == '__main__':
tf.config.list_physical_devices('GPU')
args, _ = parse_args()
FLAGS['process_max_epoch'] = args.epochs
FLAGS['folder_model']= args.model_dir
FLAGS['gpu']= args.n_gpu
FLAGS['folder_input']= args.train_input
FLAGS['folder_label']= args.label_input
tools.resize_image(FLAGS['folder_input'],max_length = 512)
tools.resize_image(FLAGS['folder_label'],max_length = 512)
tf.compat.v1.disable_eager_execution()
print(tf.config.list_physical_devices('GPU'))
#tf.compat.v1.disable_resource_variables()
os.environ["CUDA_VISIBLE_DEVICES"] = FLAGS['num_gpu']
sys.stdout = Tee(sys.stdout, open(FLAGS['txt_log'], 'a+'))
LOG_STAMP_TRAIN = np.int32(np.linspace(0, FLAGS['data_train_batch_count']-1, num=FLAGS['process_train_log_interval_epoch']+1))
LOG_STAMP_TEST = np.int32(np.linspace(0, FLAGS['data_train_batch_count']-1, num=FLAGS['process_test_log_interval_epoch'] +1))
def netG_concat_value(tensor, v):
v_t = tf.constant(v, dtype=tf.float32, shape=tensor.get_shape().as_list()[:3] + [1])
tensor = tf.concat(3, [tensor, v_t])
return tensor