* OpenVINO installation
* Separate tf_annotation -> tf_annotation and cuda support.
* TF Annotation app now supports openVino backend
* Doc for CUDA component
* OpenVINO Readme file was added
* OpenVINO env and pip requirements for model optimizer
* Update logging
* TF annotation Readme file was added
* Update CHANGELOG
* Keep aspect ratio for image, not reverse input channels
* Move analytics into components
* Incremental collection update
* CHANGELOG.md has been updated
* Detach menu container only if frame has been changed. It strongly accelerate the onCollectionUpdate method during you add objects
Nikita Manovich
Boris Sekachev