Release Notes#
Version v.4.1.0#
-
API version 6 support was added for sending data from FaceStream to LUNA PLATFORM 5.
The description of sending data to the LUNA PLATFORM was updated in the section "Image Sending to LUNA PLATFORM". This section lists the FaceStream parameters for sending images:
-
directly to LUNA PLATFORM 5,
-
to Backport 3 (implements support for LUNA PLATFORM 3 API requests in LUNA PLATFORM 5),
-
to Backport 4 (implements support for LUNA PLATFORM 4 API requests in LUNA PLATFORM 5).
-
-
Changes to input.json:
-
The "luna_account_id" parameter was moved to input.json and now available per video stream.
-
The "image_store_url" parameter was added. It enables to set the URL for the Image Store service to send the source frame to LUNA PLATFORM.
It is required to enable the "Send-source-frame" parameter to send the source frame.
The bucket for saving the source frame should be created in advance in the Image Store.
-
-
Changes to fs3config.conf:
-
The "jpeg-quality-level" setting was added. It enables you to set the quality of the original image when it is sent to the LUNA PLATFORM. By default, the image compression is not performed, which may affect the processing speed.
-
The list of tags for logging information was expanded. See the "Tags" section. The following tags were added:
-
common - general information;
-
client - information about sending messages to LUNA PLATFORM and the responses received;
-
json - information about processing parameters from configuration files and the Configurator service;
-
http_api - information about API requests sent to FaceStream in server mode and received responses;
-
image - information about frames processing;
-
gstreamer - information about GStream library operation;
-
debug - debug information. It is recommended to use this tag when debugging only and not during FS operation. It provides a large amount of debugging information.
Logs text includes the corresponding tag. It can be used for logs filtration.
-
-
-
Adding a new video stream (POST "/streams") is now performed asynchronously in server mode. The response to the request is returned immediately, without waiting for the stream to be created.
-
The FaceStream performance when using GPU was improved.
-
The stability of FaceStream when using CPU and GPU was improved.
-
The description in the "Liveness" section was extended. The requirements for the camera location, recommendations for FaceStream settings, and answers to frequently asked questions were added.
-
The "stream-images-buffer-max-size" parameter default value was set to 40 for maximum FaceStream performance. It is important to note that this leads to a significant increase in memory consumption.
Version 4.0.0#
- CPU memory leaks were fixed.
- REST API bugs were fixed.
- Logging was improved. The bug with logging severity in TrackEngine config was fixed. Added was a new severity level "debug".
- New parameters were added in TrackEngine config. Removed the parameter "fragment-buffer-size". The parameter "frame-buffer-size" renamed to "frames-buffer-size".
-
Performance and memory consumption optimizations were performed. CPU utilization was improved.
-
The "Memory consumption when running FaceStream" section was added. It describes the reasons for increasing memory consumption when running FaceStream;
- The "General recommendations for FaceStream configuration" section was added. This section lists the main configurable parameters, their impact on performance, and gives answers to other frequently asked questions.
- The "Trackengine configuration" section was added. It contains the description of TrackEngine parameters.
-
The licensing system was updated:
-
The "License activation" section was added;
-
The Sentinel HASP utility was deleted from the distribution package;
-
-
The "Use FaceStream with Configurator" section was added. The section includes specified data about the interaction of FaceStream and the Configurator service;
-
The description for adding configurations in Configurator provided in the Docker container was added;
-
Examples of JSON files with FaceStream parameters were added to the distribution package:
-
luna_configurator_dump.json,
-
luna_configurator_dump_with_streams_example.json;
-
-
-
The section with information about formats, video compression standards, and protocols used by FaceStream was added;
-
The possibility of GPU processing was implemented. Now all the video stream processing can be performed using the GPU.
-
The requirements for video cards were updated in the "System requirements";
-
The "enable-gpu-processing" parameter that enables video card utilization was added;
-
-
The new "statistics" tag to the "Tags" section was added. The tag displays information about performance, FPS, and missed frames;
- The possibility to set the number of frames processed per second using the "real_time_mode_fps" parameter was added;
- The "convert-full-frame" and "frame-processing-mode" parameters were added. They enable you to convert a decoded frame to an RGB image of the desired size, which improves the image quality;
- The "aggregate-attr-requests" parameter was added. It enables aggregation when executing the "Generate events" request in LUNA PLATFORM 4;
- The list of installed dependencies for FaceStream was updated;
- The section about installing TBB was deleted. now you don't need to install the library separately;
- The requirement for the "Name" field in input.json was added. It must be limited to 20 characters;
- The "mouth-occlusion-threshold" parameter description was updated.
- The "stream-images-buffer-max-size" parameter default value was set to 20.