ParrotLabel Scripts

Utility scripts for use with the ParrotLabel graphical image annotation tool. For creating training data for object detection and other machine learning applications.

parrotlabel_to_tfrecords

A script to convert the native ParrotLabel JSON format into the TensorFlow TFRecord format. Also produces an associated Label Map (pbtxt) file. These files can be used as an input to the TensorFlow Object Detection API for training.

Usage Example:

• Split a ParrotLabel JSON annotations file into 80% training and 20% validation (randomising the order)
• Writes the training image annontations to a TensorFlow TFRecord training file at "training_file.records"
• Writes the validation image annontations to a TensorFlow TFRecord training file at "validation.records"
• Writes a TensorFlow Label Map file to "labelmap.pbtxt"

 ./parrotlabel_to_tfrecords.py \
			--tfrecords training_file.records \
			--val_tfrecords validation.records \
			--validation_percentage 20 \
			--labelmap labelmap.pbtxt \
			--images images_dir/ \
			input_parrot_annotations.json

Options:

• --tfrecords - The file name to output the TensorFlow Records training file to
• --val_tfrecords - Optionally a file to write out a proportion of the annotations as validation examples (also in TFRecords format)
• --validation_percentage - What percentage of the annotations to use as validation example (default is 20%)
• --labelmap - The file name at which to output the associated TensorFlow label map file (pbtxt)
• --images - The path to the directory where the images are stored (use only JPEG or PNG images)

Followed by one or more ParrotLabel JSON format annotation files as arguments.

example_annotations

An example ParrotLabel JSON annotations file for testing scripts. Includes several annotated images of parrots.