Skip to content

Release 2021.2

Choose a tag to compare

View all tags
@alekseyfa alekseyfa released this 17 Dec 15:10
· 361 commits to master since this release
v1.3
a644d52
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Verified
Learn about vigilant mode.

What’s New in This Release:

Title High-level description
Direct ONNX model support DL Streamer gvadetect, gvaclassify, and gvainference elements will now support the following ONNX models supported by OpenVINO™ Inference Engine on CPU without converting to Intermediate Representation (IR) format: YOLOv2 Tiny YOLOv2 MobileNet FER+ Emotion Recognition
Support for full-frame and ROI based inference in inference elements A new property 'inference-region' added to gvadetect, gvaclassify, and gvainference element will allow developers to run object detection on ROI (Region of Interest) and object classification on full frame. This will enable use cases such as cascading of two object detection models where second model performs detection on ROI identified by the first model.
Imageless zero-term and short-term object tracking Two new algorithms 'short-term imageless' and 'zero-term imageless' introduced in gvatrack will provide an ability to track the objects without accessing image data.
Docker file and install tutorial improvements The folder structure created with the Docker file is aligned with the Docker image released by OpenVINO™ on DockerHub*. Developers can follow the same instructions and guidelines for using DL Streamer regardless of the chosen way of distribution  (OpenVINO Installer, OpenVINO Docker image, DL Streamer Docker file, building from source).
Assets 3
Loading