Related
I use Python 2.7.13 and Tensorflow 1.3.0 on CPU.
I want to use DensNet( https://github.com/pudae/tensorflow-densenet ) for regression problem. My data contains 60000 jpeg images with 37 float labels for each image.
I saved my data into tfrecords files by:
def Read_Labels(label_path):
labels_csv = pd.read_csv(label_path)
labels = np.array(labels_csv)
return labels[:,1:]
`
def load_image(addr):
# read an image and resize to (224, 224)
img = cv2.imread(addr)
img = cv2.resize(img, (224, 224), interpolation=cv2.INTER_CUBIC)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
img = img.astype(np.float32)
return img
def Shuffle_images_with_labels(shuffle_data, photo_filenames, labels):
if shuffle_data:
c = list(zip(photo_filenames, labels))
shuffle(c)
addrs, labels = zip(*c)
return addrs, labels
def image_to_tfexample_mine(image_data, image_format, height, width, label):
return tf.train.Example(features=tf.train.Features(feature={
'image/encoded': bytes_feature(image_data),
'image/format': bytes_feature(image_format),
'image/class/label': _float_feature(label),
'image/height': int64_feature(height),
'image/width': int64_feature(width),
}))
def _convert_dataset(split_name, filenames, labels, dataset_dir):
assert split_name in ['train', 'validation']
num_per_shard = int(math.ceil(len(filenames) / float(_NUM_SHARDS)))
with tf.Graph().as_default():
for shard_id in range(_NUM_SHARDS):
output_filename = _get_dataset_filename(dataset_path, split_name, shard_id)
with tf.python_io.TFRecordWriter(output_filename) as tfrecord_writer:
start_ndx = shard_id * num_per_shard
end_ndx = min((shard_id+1) * num_per_shard, len(filenames))
for i in range(start_ndx, end_ndx):
sys.stdout.write('\r>> Converting image %d/%d shard %d' % (
i+1, len(filenames), shard_id))
sys.stdout.flush()
img = load_image(filenames[i])
image_data = tf.compat.as_bytes(img.tostring())
label = labels[i]
example = image_to_tfexample_mine(image_data, image_format, height, width, label)
# Serialize to string and write on the file
tfrecord_writer.write(example.SerializeToString())
sys.stdout.write('\n')
sys.stdout.flush()
def run(dataset_dir):
labels = Read_Labels(dataset_dir + '/training_labels.csv')
photo_filenames = _get_filenames_and_classes(dataset_dir + '/images_training')
shuffle_data = True
photo_filenames, labels = Shuffle_images_with_labels(
shuffle_data,photo_filenames, labels)
training_filenames = photo_filenames[_NUM_VALIDATION:]
training_labels = labels[_NUM_VALIDATION:]
validation_filenames = photo_filenames[:_NUM_VALIDATION]
validation_labels = labels[:_NUM_VALIDATION]
_convert_dataset('train',
training_filenames, training_labels, dataset_path)
_convert_dataset('validation',
validation_filenames, validation_labels, dataset_path)
print('\nFinished converting the Flowers dataset!')
And I decode it by:
with tf.Session() as sess:
feature = {
'image/encoded': tf.FixedLenFeature((), tf.string, default_value=''),
'image/format': tf.FixedLenFeature((), tf.string, default_value='jpeg'),
'image/class/label': tf.FixedLenFeature(
[37,], tf.float32, default_value=tf.zeros([37,], dtype=tf.float32)),
}
filename_queue = tf.train.string_input_producer([data_path], num_epochs=1)
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(serialized_example, features=feature)
image = tf.decode_raw(features['image/encoded'], tf.float32)
print(image.get_shape())
label = tf.cast(features['image/class/label'], tf.float32)
image = tf.reshape(image, [224, 224, 3])
images, labels = tf.train.shuffle_batch([image, label], batch_size=10, capacity=30, num_threads=1, min_after_dequeue=10)
init_op = tf.group(tf.global_variables_initializer(), tf.local_variables_initializer())
sess.run(init_op)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
for batch_index in range(6):
img, lbl = sess.run([images, labels])
img = img.astype(np.uint8)
print(img.shape)
for j in range(6):
plt.subplot(2, 3, j+1)
plt.imshow(img[j, ...])
plt.show()
coord.request_stop()
coord.join(threads)
It's all fine up to this point. But when I use the bellow commands for decoding TFRecord files:
reader = tf.TFRecordReader
keys_to_features = {
'image/encoded': tf.FixedLenFeature((), tf.string, default_value=''),
'image/format': tf.FixedLenFeature((), tf.string, default_value='raw'),
'image/class/label': tf.FixedLenFeature(
[37,], tf.float32, default_value=tf.zeros([37,], dtype=tf.float32)),
}
items_to_handlers = {
'image': slim.tfexample_decoder.Image('image/encoded'),
'label': slim.tfexample_decoder.Tensor('image/class/label'),
}
decoder = slim.tfexample_decoder.TFExampleDecoder(
keys_to_features, items_to_handlers)
I get the following error.
INFO:tensorflow:Error reported to Coordinator: , assertion failed: [Unable to decode bytes as JPEG, PNG, GIF, or BMP]
[[Node: case/If_0/decode_image/cond_jpeg/cond_png/cond_gif/Assert_1/Assert = Assert[T=[DT_STRING], summarize=3, _device="/job:localhost/replica:0/task:0/cpu:0"](case/If_0/decode_image/cond_jpeg/cond_png/cond_gif/is_bmp, case/If_0/decode_image/cond_jpeg/cond_png/cond_gif/Assert_1/Assert/data_0)]]
INFO:tensorflow:Caught OutOfRangeError. Stopping Training.
INFO:sensorflow:Finished training! Saving model to disk.
To use Densenet for my problem, I should fix this error first.
Could anybody please help me out of this problem. This code works perfectly for the datasets like flowers, MNIST and CIFAR10 available at https://github.com/pudae/tensorflow-densenet/tree/master/datasets but does not work for my data.
Thanks to pudae, the problem is solved. I was needed to use:
image_data = tf.gfile.FastGFile(filenames[i], 'rb').read()
Instead of this for loading data. That works perfectly now.
img = load_image(filenames[i])
image_data = tf.compat.as_bytes(img.tostring())
According to the error, I think the problem is that you use an image decoder for array data (decoded data) because you saved decoded data when creating TFRecords. Maybe you have noticed, when you are not using slim, you use tf.decode_raw to decode the data. But when you use slim, the 'image/format': tf.FixedLenFeature((), tf.string, default_value='raw') is not used and by default, slim will use image decoder.
I believe you use the code in slim/data,
where format_key = 'image/format' is you need. So, like this:
keys_to_features = {
'image/encoded': tf.FixedLenFeature((), tf.string, default_value=''),
'image/format': tf.FixedLenFeature((), tf.string, default_value='raw'),
'image/class/label': tf.FixedLenFeature(
[1], tf.int64, default_value=tf.zeros([1], dtype=tf.int64)),
}
items_to_handlers = {
'image': tfexample_decoder.Image(
image_key = 'image/encoded',
format_key = 'image/format',
'label': tfexample_decoder.Tensor('image/class/label'),
}
decoder = tfexample_decoder.TFExampleDecoder(
keys_to_features, items_to_handlers)
But I am not sure this can solve your problem perfectly because I can't reproduce your work in my machine.
Maybe there is a problem with your image itself as follows：
I'm a Python and Tensor Flow newbie, and was wondering...
How best to convert a labelled dataset of Multi-Layer Tiffs into a format that Tensor Flow can use for model optimisation / fine tuning ?
I currently have this code that puts each layer of a folder of Multi-Tiffs into a 3D Array, but i need to preserve the label or filename of the Multi-Tiffs. I have seen some tensor flow scripts to convert to TFRecords, however, I'm not sure if these preserve the file name ? How best would you go about this ? It will be quite a big dataset.
Any help much appreciated
import os # For file handling
from PIL import Image# Import Pillow image processing library
import numpy
CroppedMultiTiffs = "MultiTiffs/"
for filename in os.listdir(MultiTiffs):
## Imports Multi-Layer TIFF into 3D Numpy Array.
img = Image.open(MultiTiffs + filename)
imgArray = numpy.zeros( ( img.n_frames, img.size[1], img.size[0] ),numpy.uint8 )
try:
# for frames in range, img.n_frames for whole folder.
for frame in range(2,img.n_frames):
img.seek( frame )
imgArray[frame,:,:] = img
frame = frame + 1
except (EOFError): img.seek( 0 )
# output error if it doesn't find a file.
pass
print(imgArray.shape) # imgArray is now 3D
print(imgArray.size)
best wishes
TWP
okay, so I figured it out using the thread from Daniils blog
http://warmspringwinds.github.io/tensorflow/tf-slim/2016/12/21/tfrecords-guide/
However my current implimentation creates multiple TFRecords, and I think it needs to be a single TFRecord, so trying to figure out how to make it a single TFRecord. How do I do that?
Then I can validate it using a TFRecord Reading script to read it back and check it is in the right format for Tensor Flow. I currently get errors using the reading script.
from PIL import Image
import numpy as np
import tensorflow as tf
import os
def _bytes_feature(value):
return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
def _int64_feature(value):
return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))
path = 'test/'
output = 'output/'
fileList = [os.path.join(dirpath, f) for dirpath, dirnames, files in os.walk(path) for f in files if f.endswith('.tif')]
print (fileList)
for filename in fileList:
basename = os.path.basename(filename)
file_name = basename[:-4]
print ("processing file: " , filename)
print (file_name)
if not os.path.exists(output):
os.mkdir(output)
writer = tf.python_io.TFRecordWriter(output+ file_name + '.tfrecord')
img = Image.open(filename)
imgArray = np.zeros( ( img.n_frames, img.size[1], img.size[0] ),np.uint8 )
## Imports Multi-Layer file into 3D Numpy Array.
try:
for frame in range(0,img.n_frames):
img.seek( frame )
imgArray[frame,:,:] = img
frame = frame + 1
except (EOFError): img.seek( 0 )
pass
print ("print img size:" , img.size)
print ("print image shape: " , imgArray.shape)
print ("print image size: " , imgArray.size)
annotation = np.array(Image.open(filename))
height = imgArray.shape[0]
width = imgArray.shape[1]
depth = imgArray.shape[2]
img_raw = imgArray.tostring()
annotation_raw = annotation.tostring()
example = tf.train.Example(features=tf.train.Features(feature={
'height': _int64_feature(height),
'width': _int64_feature(width),
'depth': _int64_feature(depth), # for 3rd dimension
'image_raw': _bytes_feature(img_raw),
'mask_raw': _bytes_feature(annotation_raw)}))
writer.write(example.SerializeToString())
My current TFRecords Reading script
import tensorflow as tf
import os
def read_and_decode(filename_queue):
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'image_raw': tf.FixedLenFeature([], tf.string),
'label': tf.FixedLenFeature([], tf.int64),
'height': tf.FixedLenFeature([], tf.int64),
'width': tf.FixedLenFeature([], tf.int64),
'depth': tf.FixedLenFeature([], tf.int64)
})
image = tf.decode_raw(features['image_raw'], tf.uint8)
label = tf.cast(features['label'], tf.int32)
height = tf.cast(features['height'], tf.int32)
width = tf.cast(features['width'], tf.int32)
depth = tf.cast(features['depth'], tf.int32)
return image, label, height, width, depth
with tf.Session() as sess:
filename_queue = tf.train.string_input_producer(["output/A.3.1.tfrecord"])
image, label, height, width, depth = read_and_decode(filename_queue)
image = tf.reshape(image, tf.stack([height, width, 3]))
image.set_shape([32,32,3])
init_op = tf.initialize_all_variables()
sess.run(init_op)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
for i in range(1000):
example, l = sess.run([image, label])
print (example,l)
coord.request_stop()
coord.join(threads)
receiving the error:-
InvalidArgumentError (see above for traceback): Name: , Feature: label (data type: int64) is required but could not be found.
Images are grayscale multi-page
I'm using a script that comes with TF-Slim to validate my trained model. It works fine but I'd like to get a list of the misclassified files.
The script makes use of https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/slim/python/slim/evaluation.py but even there I cannot find any options for printing the misclassified files.
How can I achieve that?
At a high level, you need to do 3 things:
1) Get your filename from the data loader. If you are using a tf-slim dataset from tfrecords, it is likely that the filenames are not stored in the tfrecord so you may be out of luck there. However if you are consuming image files directly from the filesystem with tf.WholeFileReader, then you can get the tensor of filenames where you form your batch:
def load_data():
train_image_names = ... # list of filenames
filename_queue = tf.train.string_input_producer(train_image_names)
reader = tf.WholeFileReader()
image_filename, image_file = reader.read(filename_queue)
image = tf.image.decode_jpeg(image_file, channels=3)
.... # load your labels from somewhere
return image_filename, image, label
# in your eval code
image_fn, image, label = load_data()
filenames, images, labels = tf.train.batch(
[image_fn, image, label],
batch_size=32,
num_threads=2,
capacity=100,
allow_smaller_final_batch=True)
2) Mask your filename tensor with your result after doing inference:
logits = my_network(images)
preds = tf.argmax(logits, 1)
mislabeled = tf.not_equal(preds, labels)
mislabeled_filenames = tf.boolean_mask(filenames, mislabeled)
3) Put all this into your eval_op:
eval_op = tf.Print(eval_op, [mislabeled_filenames])
slim.evaluation.evaluate_once(
.... # other options
eval_op=eval_op,
.... # other options)
I don't have a setup to test this, unfortunately. Let me know if it works!
shadow chris pointed me in the right direction so I share my solution to make it work with a TF-records dataset.
For better unstanding I relate my code to the flower example of TF-Slim.
1) Modify your dataset script to store a filename feature in the TF-records.
keys_to_features = {
'image/encoded': tf.FixedLenFeature((), tf.string, default_value=''),
'image/format': tf.FixedLenFeature((), tf.string, default_value='png'),
'image/class/label': tf.FixedLenFeature(
[], tf.int64, default_value=tf.zeros([], dtype=tf.int64)),
'image/filename': tf.FixedLenFeature((), tf.string, default_value=''),
}
items_to_handlers = {
'image': slim.tfexample_decoder.Image(),
'label': slim.tfexample_decoder.Tensor('image/class/label'),
'filename': slim.tfexample_decoder.Tensor('image/filename'),
}
2) Add filename parameter to data util's image_to_tfexample function
It should then look like:
def image_to_tfexample(image_data, image_format, height, width, class_id, filename):
return tf.train.Example(features=tf.train.Features(feature={
'image/encoded': bytes_feature(image_data),
'image/format': bytes_feature(image_format),
'image/class/label': int64_feature(class_id),
'image/height': int64_feature(height),
'image/width': int64_feature(width),
'image/filename': bytes_feature(filename)
}))
3) Modify download and convert script to save filenames
Feed your TF record with the filename.
example = dataset_utils.image_to_tfexample(
image_data, 'jpg', height, width, class_id, filenames[i])
4) In your evaluation map misclassified imgs to filename
I'm refering to eval_image_classifier.py.
Retrieve filenames with tf.train.batch:
images, labels, filenames = tf.train.batch(
[image, label, filename],
batch_size=FLAGS.batch_size,
num_threads=FLAGS.num_preprocessing_threads,
capacity=5 * FLAGS.batch_size)
Get misclassified imgs and map them to filenames:
predictions = tf.argmax(logits, 1)
labels = tf.squeeze(labels)
mislabeled = tf.not_equal(predictions, labels)
mislabeled_filenames = tf.boolean_mask(filenames, mislabeled)
Print:
eval_op = tf.Print(eval_op, [mislabeled_filenames])
slim.evaluation.evaluate_once(
.... # other options
eval_op=eval_op,
.... # other options)
I have converted a directory of images and their labels into a TFRecords file, the feature maps include image_raw, label, height, width and depth. The function is as follows:
def convert_to_tfrecords(data_samples, filename):
def _int64_feature(value):
return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))
def _bytes_feature(value):
return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
writer = tf.python_io.TFRecordWriter(filename)
for fname, lb in data_samples:
im = cv2.imread(fname, cv2.IMREAD_UNCHANGED)
image_raw = im.tostring()
feats = tf.train.Features(
feature =
{
'image_raw': _bytes_feature(image_raw),
'label': _int64_feature(int(lb)),
'height': _int64_feature(im.shape[0]),
'width': _int64_feature(im.shape[1]),
'depth': _int64_feature(im.shape[2])
}
)
example = tf.train.Example(features=feats)
writer.write(example.SerializeToString())
writer.close()
Now, I would like to read this TFRecords file to feed a input pipeline. However, since image_raw has been flattened, we need to reshape it into the original [height, width, depth] size. So how can I get the values of height, width and depth from the TFRecords file? It seems the following code cannot work because height is a Tensor without values.
def read_and_decode(filename_queue):
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
feats = {
'image_raw': tf.FixedLenFeature([], tf.string),
'label': tf.FixedLenFeature([], tf.int64),
'height': tf.FixedLenFeature([], tf.int64),
'width': tf.FixedLenFeature([], tf.int64),
'depth': tf.FixedLenFeature([], tf.int64)
}
features = tf.parse_single_example(serialized_example, features=feats)
image = tf.decode_raw(features['image_raw'], tf.uint8)
label = tf.cast(features['label'], tf.int32)
height = tf.cast(features['height'], tf.int32)
width = tf.cast(features['width'], tf.int32)
depth = tf.cast(features['depth'], tf.int32)
image = tf.reshape(image, [height, width, depth]) # <== not work
image = tf.cast(image, tf.float32) * (1. / 255) - 0.5
return image, label
When I read the Tensorflow's official documents, I found they usually pass into a known size, saying [224,224,3]. However, I don't like it, because this information has been stored into the TFRecords file, and manually passing into fixed size cannot ensure the size is consistent with the data stored in the file.
So any ideas?
The height returned by tf.parse_single_example is a Tensor, and the only way to get its value is to call session.run() on it, or similar. However, I think that's overkill.
Since the Tensorflow example is just a protocol buffer (see the documentation), you don't necessarily have to use tf.parse_single_example to read it. You could instead parse it yourself and read the shapes you want out directly.
You might also consider filing a feature request on Tensorflow's github issues tracker --- I agree this API seems a bit awkward for this use case.
The function 'tf.reshape' only accept a tensor,not a list of tensors,so you can use the following code:
image = tf.reshape(image, tf.stack([height, width, depth]))
You can also get the numpy array out ot the tensor and reshape using np.resize() passing the dimensions as argument
The main idea is to convert TFRecords into numpy arrays. Assume that the TFRecord stores images. Specifically:
Read a TFRecord File and convert each image into a numpy array.
Write the image into 1.jpg, 2.jpg, etc.
At the same time, write the file name and label to the text file like this:
1.jpg 2
2.jpg 4
3.jpg 5
I currently use the following code:
import tensorflow as tf
import os
def read_and_decode(filename_queue):
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'image_raw': tf.FixedLenFeature([], tf.string),
'label': tf.FixedLenFeature([], tf.int64),
'height': tf.FixedLenFeature([], tf.int64),
'width': tf.FixedLenFeature([], tf.int64),
'depth': tf.FixedLenFeature([], tf.int64)
})
image = tf.decode_raw(features['image_raw'], tf.uint8)
label = tf.cast(features['label'], tf.int32)
height = tf.cast(features['height'], tf.int32)
width = tf.cast(features['width'], tf.int32)
depth = tf.cast(features['depth'], tf.int32)
return image, label, height, width, depth
with tf.Session() as sess:
filename_queue = tf.train.string_input_producer(["../data/svhn/svhn_train.tfrecords"])
image, label, height, width, depth = read_and_decode(filename_queue)
image = tf.reshape(image, tf.pack([height, width, 3]))
image.set_shape([32,32,3])
init_op = tf.initialize_all_variables()
sess.run(init_op)
print (image.eval())
I'm just reading trying to get at least one image for starters. The code just gets stuck when I run this.
Oops, it was a silly mistake on my part. I used a string_input_producer but forgot to run the queue_runners.
with tf.Session() as sess:
filename_queue = tf.train.string_input_producer(["../data/svhn/svhn_train.tfrecords"])
image, label, height, width, depth = read_and_decode(filename_queue)
image = tf.reshape(image, tf.pack([height, width, 3]))
image.set_shape([32,32,3])
init_op = tf.initialize_all_variables()
sess.run(init_op)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
for i in range(1000):
example, l = sess.run([image, label])
print (example,l)
coord.request_stop()
coord.join(threads)