Skip to content

Latest commit

 

History

History
37 lines (28 loc) · 1.94 KB

ReleaseNotes.md

File metadata and controls

37 lines (28 loc) · 1.94 KB

iVSR Release Notes

Next Release

New and Changes

Bug Fixes

Known Limitations/Issues

  • If the model-guard protected model is loaded, it prints the following error messages. They can be ignored as its normal procedure for this kind of model files.
    [libprotobuf ERROR thirdparty/protobuf/protobuf/src/google/protobuf/text_format.cc:335] Error parsing text-format tensorflow.GraphDef: 1:2: Message type "tensorflow.GraphDef" has no field named "T".
    [libprotobuf ERROR thirdparty/protobuf/protobuf/src/google/protobuf/text_format.cc:335] Error parsing text-format tensorflow.GraphDef: 1:2: Message type "tensorflow.GraphDef" has no field named "T".

Release v24.05

New and Changes in v24.05

  • Updated iVSR plugin to FFmpeg n6.1.
  • Enabled the ability to change model input shapes in the SDK.
  • Enabled Y-frame input support.
  • Updated Enhanced BasicVSR and Enhanced EDSR models to support reshape.
  • Updated SVP-Basic and SVP-SE models with improved inference performance.
  • Preview version of TSENet model support.
  • Removed the "input_size" parameter from the ivsr_process() API.
  • Added one build option to disable applying OpenVINO patches which are required to support Enhanced BasicVSR model.

Bug Fixes

  • Fixed the issue of improper calling of ivsr_process() for the first frame in the FFmpeg plugin.
  • Refined the SDK initialization process in the FFmpeg plugin, and removed several misleading warnings.

Known Limitations

  • iVSR SDK is runing in sync mode only. Aync mode is not enabled yet.
  • The patch-based solution feature is not enabled in the FFmpeg plugin yet. It can be enabled by setting vsr_sample's "reshape_values" option.
  • Parallel inference on heterogeneous hardware is not enabled yet.
  • Only 8-bit data is supported in vsr_sample or iVSR FFmpeg plugin.
  • Enhanced BasicVSR model works only on OpenVINO 2022.3 with the patches applied.
  • The Enhanced BasicVSR model supports only input sizes with widths that are divisible by 32.