CHANGELOG.md

Changelog

All notable changes to this project will be documented in this file.

This project is based on MMDetection project by OpenMMLab.

With respect to it we made the following changes.

[2022-05-09]

Added

• Object counting models using Mask-RCNN
• Rotated object detection models using Mask-RCNN
• YoloX with NNCF support
• Stop training on NaN losses
• FP16 support

Removed

• Tasks & model templates (moved to OTE)

[2021-12-27]

Added

• OTE SDK Task interface implementation
• OpenVINO export including alternative export for SSD models
• Models, datasets, metrics for text detection and recognition
• Backbones from pytorchcv and EfficientDet architecture
• Model compression possibility by NNCF and config examples
• Configuration files for
  • custom object detection
  • face detection
  • horizontal text detection
  • person detection
  • person-vehicle-bike detection
  • vehicle detection
• Extra augmentstions and transforms, ReduceOnPlateau LR scheduler, early stopping hook
• Extra tool scripts for
  • COCO dataset visualization and filtration
  • clustering anchor boxes for SSD anchor generator
  • validation of the ONNX and OpenVINO models

Changed

• Improved the export to the ONNX including runtime mode
• Added OpenVINO installation to the dockerfile configuration