def get_batch_of_stored_bottlenecks(sess, image_lists, batch_size, category,
bottleneck_dir, image_dir,
jpeg_data_tensor, decoded_image_tensor,
resized_input_tensor, bottleneck_tensor):
class_count = len(image_lists.keys())
bottlenecks = []
ground_truths = []
filenames = []
if batch_size >= 0:
# Retrieve a random sample of bottlenecks.
for i in range(batch_size):
label_index = random.randrange(class_count)
label_name = list(image_lists.keys())[label_index]
image_index = random.randrange(MAX_NUM_IMAGES_PER_CLASS + 1)
image_name = utils.get_image_path(image_lists, label_name,
image_index, image_dir, category)
bottleneck = get_bottleneck(sess, image_lists, label_name,
image_index, image_dir, category,
bottleneck_dir, jpeg_data_tensor,
decoded_image_tensor,
resized_input_tensor,
bottleneck_tensor)
ground_truth = np.zeros(class_count, dtype=np.float32)
ground_truth[label_index] = 1.0
bottlenecks.append(bottleneck)
ground_truths.append(ground_truth)
filenames.append(image_name)
else:
# Retrieve all bottlenecks
# Used for validation set mainly
for label_index, label_name in enumerate(image_lists.keys()):
for image_index, image_name in enumerate(
image_lists[label_name][category]):
image_name = utils.get_image_path(image_lists, label_name,
image_index, image_dir,
category)
bottleneck = get_bottleneck(sess, image_lists, label_name,
image_index, image_dir, category,
bottleneck_dir, jpeg_data_tensor,
decoded_image_tensor,
resized_input_tensor,
bottleneck_tensor)
ground_truth = np.zeros(class_count, dtype=np.float32)
ground_truth[label_index] = 1.0
bottlenecks.append(bottleneck)
ground_truths.append(ground_truth)
filenames.append(image_name)
return bottlenecks, ground_truths, filenames
def create_bottleneck(bottleneck_path, image_lists, label_name, index,
image_dir, category, sess, jpeg_data_tensor,
decoded_image_tensor, resized_input_tensor,
bottleneck_tensor):
tf.logging.info("Creating Bottleneck at {}".format(bottleneck_path))
image_path = utils.get_image_path(image_lists, label_name, index,
image_dir, category)
if not gfile.Exists(image_path):
tf.logging.fatal("File does not exist {}".format(image_path))
image_data = gfile.FastGFile(image_path, "rb").read()
try:
bottleneck_values = run_bottleneck_on_image(sess, image_data,
jpeg_data_tensor,
decoded_image_tensor,
resized_input_tensor,
bottleneck_tensor)
except Exception as e:
raise RuntimeError("Error bottlenecking {}\n{}".format(
image_path, str(e)))
bottleneck_string = ",".join(str(x) for x in bottleneck_values)
bottleneck_directory = "/".join(bottleneck_path.split("/")[:-1])
utils.create_directory(bottleneck_directory)
with open(bottleneck_path, "w") as bottleneck_file:
bottleneck_file.write(bottleneck_string)