Release notes version 00.05.00

New in this Release

  • Python and C++ demos supported on both docker and host

  • ONNX Runtime support

  • RTSP video stream support

  • Single input, multiple inference support

  • Multi input, multiple inference support

Fixed in this Release

  • ADASVISION-4719 - C++ apps exits throwing an error for video inputs

  • ADASVISION-4723 - Python demos cannot be run in docker session

  • ADASVISION-4724 - C++ demos cannot be built or run in the host OS

  • ADASVISION-4780 - Demo apps hang while running them back to back

  • ADASVISION-4784 - Docker run command not spawning to /opt/edge_ai_apps

  • ADASVISION-4785 - C++ application crashes for TVM based mxnet and video input

  • ADASVISION-4790 - ONNX semantic segmentation model output corrupted

  • ADASVISION-4791 - Few regNetx ONNX & TVM classification models are not running in Python demo

  • ADASVISION-4792 - Python demos performance poorer than C++ demos on 0.5 RC

  • ADASVISION-4793 - Performance of demos in 0.5RC poorer than 0.4.2

  • ADASVISION-4798 - Font size of Top5 class is large for multichannel display

Known Issues

  • ADASVISION-4720 - Output video encode framerate incorrect it is hardcoded to 30fps

  • ADASVISION-4721 - Docker session restart required sometimes when demos exit abruptly with Ctrl+C

  • ADASVISION-4722 - Demo with video input does not loop continuously like image inputs

  • ADASVISION-4781 - install_dl_dependencies.sh is not working on Host OS

  • ADASVISION-4783 - Performance issues observed for more than 2 USB cameras

  • ADASVISION-4794 - Application tasks 10-15 seconds to come up in some cases

  • ADASVISION-4795 - Some ONNX classification models occasionally do not run

  • ADASVISION-4796 - Frame drop observed with video input

  • ADASVISION-4805 - Default etcher image size not sufficient for docker build and all model download

  • ADASVISION-4806 - Python based Multi-input, Multi-output demo stops after video stops

  • ADASVISION-4807 - Model name font size does not scale

  • ADASVISION-4809 - Python and C++ apps cannot measure pre and post processing time accurately

Software components

List of software components used in this version

Componenet

Version

PSDKLA (filesystem, uboot)

07.03.01

PSDKRA (firmware)

07.03.01

Docker

19.03.8-ce

Ubuntu (Docker)

20.04.LTS

Python

3.8

OpenCV

4.5.1

GStreamer

1.16.2

Cmake

3.16.3

Ninja

1.10.0

Meson

0.57.2

Jupyterlab

3.0.14

NeoAI - DLR

1.8.0

Tensorflow

2.4

TFLite-runtime

2.5.0

ONNX-runtime

1.7.0

PyYAML

5.4.1

TI Model Zoo

2.0