realtime_object_detection 2.1

This release removes all ancient relicts of the original TF object_detection API.
The stuff necessary like label_map_util.py and some mask transformation are gathered in tf_utils.py.

New or refurbished modules:

• vis_utils.py
• ros.py
• tf_utils.py

realtime_object_detection 2.0

This release merges the realtime_object detection v1 and realtime_segmentation projects.
It includes the following features:

• all development that went into v1 (split_model speed hack etc.)
• usage of all research/object_detection and research/deeplab models
• scripts for statistical inference test runs on sample /test_images
• scripts to summarize, inspect and transform all models

realtime_object_detection 1.0

This release is the final version of the original object detection repo as it was initially designed.

• It is capable of detecting boundingboxes and masks for all tf object_detection models.
• It includes the split_model and multithreading speed hacks for ssd_mobilenet models.
• This release contains the old directory structure, newer releases will include deeplab models and will therefore be completely restructured