Segmentation of Machine Readable Zones using synthetic data generation

This notebook shows a proof of concept on doing segmentation of the left angle brackets in the machine readable zone (mrz) of passports, using synthetic data generation.

References

• Arlazarov et al. - MIDV-500: A Dataset for Identity Documents Analysis and Recognition on Mobile Devices in Video Stream
• fcakyon/midv500: Download and convert MIDV-500 annotations to COCO instance segmentation format
• qubvel/segmentation_models: Segmentation models with pretrained backbones. Keras and TensorFlow Keras.
• tensorflow/tensorflow: An Open Source Machine Learning Framework for Everyone
• python-pillow/Pillow: The friendly PIL fork (Python Imaging Library)