I'm quit new at Python, and tried to launch the script
convert_to_tfrecord.py
(Neural Networks; It should train dataset of Images with some libraries ...)
Instruction:
Now you're ready to run the TFRecord script. Run the command below from the tensorflow/models/research directory, and pass it the following flags (run it twice: once for training data, once for test data):
python convert_labels_to_tfrecords.py \
--output_path=train.record \
--images_dir=path/to/your/training/images/ \
--labels_dir=path/to/training/label/xml/
So to suit my OS X, I ran this script by python3... changes name of script... and set a directory ... I'm at directory where my script is; Where my folder is; Where my libraries are. So in my case:
python3 convert_to_tfrecord.py \
--output_path=train.record \
--images_dir=ENFj/ \
--labels_dir=ENFj/xml/
Result:
Oleksandrs-MacBook-Air:research jaskier$ python3 convert_to_tfrecord.py \
> --output_path=train.record \
Traceback (most recent call last):
File "convert_to_tfrecord.py", line 89, in <module>
tf.app.run()
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 126, in run
_sys.exit(main(argv))
File "convert_to_tfrecord.py", line 81, in main
for filename in os.listdir(FLAGS.images_dir):
FileNotFoundError: [Errno 2] No such file or directory: ''
Oleksandrs-MacBook-Air:research jaskier$ --images_dir=ENFj/ \
> --labels_dir=ENFj/xml/
Code of "convert_to_tfrecord":
import os
import io
import xml.etree.ElementTree as ET
import tensorflow as tf
from object_detection.utils import dataset_util
from PIL import Image
flags = tf.app.flags
flags.DEFINE_string('output_path', '', 'Path to output TFRecord')
flags.DEFINE_string('images_dir', '', 'Path to directory of images')
flags.DEFINE_string('labels_dir', '', 'Path to directory of labels')
FLAGS = flags.FLAGS
def create_tf_example(example):
image_path = os.getcwd() + '/' + FLAGS.images_dir + example
labels_path = os.getcwd() + '/' + FLAGS.labels_dir + os.path.splitext(example)[0] + '.xml'
# Read the image
img = Image.open(image_path)
width, height = img.size
img_bytes = io.BytesIO()
img.save(img_bytes, format=img.format)
height = height
width = width
encoded_image_data = img_bytes.getvalue()
image_format = img.format.encode('utf-8')
# Read the label XML
tree = ET.parse(labels_path)
root = tree.getroot()
xmins = xmaxs = ymins = ymaxs = list()
for coordinate in root.find('object').iter('bndbox'):
xmins = [int(coordinate.find('xmin').text)]
xmaxs = [int(coordinate.find('xmax').text)]
ymins = [int(coordinate.find('ymin').text)]
ymaxs = [int(coordinate.find('ymax').text)]
classes_text = ['tswift'.encode('utf-8')]
classes = [1]
tf_example = tf.train.Example(features=tf.train.Features(feature={
'image/height': dataset_util.int64_feature(height),
'image/width': dataset_util.int64_feature(width),
'image/filename': dataset_util.bytes_feature(encoded_image_data),
'image/source_id': dataset_util.bytes_feature(encoded_image_data),
'image/encoded': dataset_util.bytes_feature(encoded_image_data),
'image/format': dataset_util.bytes_feature(image_format),
'image/object/bbox/xmin': dataset_util.float_list_feature(xmins),
'image/object/bbox/xmax': dataset_util.float_list_feature(xmaxs),
'image/object/bbox/ymin': dataset_util.float_list_feature(ymins),
'image/object/bbox/ymax': dataset_util.float_list_feature(ymaxs),
'image/object/class/text': dataset_util.bytes_list_feature(classes_text),
'image/object/class/label': dataset_util.int64_list_feature(classes),
}))
return tf_example
def main(_):
writer = tf.python_io.TFRecordWriter(FLAGS.output_path)
for filename in os.listdir(FLAGS.images_dir):
tf_example = create_tf_example(filename)
writer.write(tf_example.SerializeToString())
writer.close()
if __name__ == '__main__':
tf.app.run()
Tried different stuff: 1. UPDATED request to (one line + deleted "\\" because as someone mentioned below, it was an error of page's PHP interpreter... it used \\n and forgot to hide the output text):
python3 convert_to_tfrecord.py
--output_path=train.record
--images_dir=ENFj
--labels_dir=ENFj/xml
2."find . -name ".DS_Store" -delete" 3. Tried again:
File "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/PIL/Image.py", line 2543, in open fp = builtins.open(filename, "rb") IsADirectoryError: [Errno 21] Is a directory: '/Users/jaskier/Downloads/models/research/ENFj/xml'
The code is set up to interpret all files in --images_dir
to some image processing machinery. This means that any non-image resources in the --images_dir
will cause the script to break.
One solution is to ensure --images_dir
contains only image files (ie ensure that directory doesn't contain XML files or files that begin with a ., like .git
or .DS_Store
.
Another solution would be to modify the source code itself to only work on image files. Something like this could be used:
import glob
# only match jpg files in the images_dir
for filename in glob.glob(FLAGS.images_dir + '/*.jpg'):
tf_example = create_tf_example(filename)
# copy the other lines here as needed
In my case it was just a simple dumb mistake: when I changed manually the name of my folder:
I have left a SPACE after the name - and when I was calling it with os or glob.glob I was getting: FileNotFoundError: [Errno 2] No such file or directory:
After spending some time searching for the problem(even changing some files names and removing some subfolders) I went back to the folder, deleted that extra space, and everything worked just fine.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.