Skip to content

v.5.141.0#

Changes

  • SDK has been updated to version 5.34.0.

  • A stream retransmission mechanism using MediaMtx has been added to the Video Agent service to extend compatibility with formats that were not previously supported (e.g., WebRTC).

    Note: This feature is currently in beta testing.

    If a stream cannot be processed using the standard method, the agent automatically attempts to retransmit it to a supported format. This functionality is not supported for the videofile stream type.

    To enable this functionality, you must:

  • A new series, agent_stream_eof, has been added to the Video Agent service monitoring, which records reaching the end of the stream (END_OF_FILE) during video processing.

    For more information, see the "Monitoring" section of the developer guide.

  • A GPU availability check has been added to the Remote SDK service's /healthcheck resource, allowing for quick identification of GPU issues.

  • Now when creating, updating and importing lambdas with the enable_gpu parameter enabled, the NVIDIA_DRIVER_CAPABILITIES environment variable with the value video,compute,utility is added to the Lambda service container.

    This allows you to fully utilize all the GPU's capabilities and avoid errors associated with incorrect driver initialization.

Fixed errors

  • Fixed an issue where the people_counter, face_quality, body_attributes, liveness, and deepfake estimation features were incorrectly reported as unavailable (false) in the response to the "get platform features" request if the Remote SDK service was disabled.

    Now feature availability is determined exclusively by the presence of the corresponding license rights, regardless of the Remote SDK service status.

  • Fixed a behavior that caused an internal server error (status code 500) when executing the save events request if the detection data contained the headwear body attribute with a valid state field, but without apparent_color specified.

    Previously, if apparent_color was missing, the value None was substituted, resulting in a validation error. Now, if the field is not specified in the request, the headwear object is generated with only the state field.

  • Fixed a validation error with the lambda_type parameter for the create lambda and put lambda requests.

    Previously, creating/updating an agent lambda using the "application/json" schema returned error 12022, despite the agent value being valid. One of the following values ​​was incorrectly required: "standalone", "handlers", or "tasks".

LUNA Vinder Module

  • Starting with this build, a new module LUNA Vinder Module (LVM) is available for LUNA PLATFORM, designed to enable flexible and high-performance event matching.

    LUNA Vinder Module is based on an architecture that separates data storage and search. Instead of working directly with the main database, the system creates specialized data copies — projections — which significantly speeds up matching and reduces the load on the primary storage. LVM allows you to:

    • Create projections — optimized data subsets with selected attributes for specific search scenarios;
    • Configure indexes to optimize the projection comparison process by defining which attributes can be used for filtering;
    • Integrate with the existing LUNA PLATFORM infrastructure via the Python Matcher Proxy matching plugin.

    Key advantages of the module:

    • High performance through the use of projections and indexes;
    • Configuration flexibility – the ability to create multiple projections for different search scenarios with individual filters and attribute sets;
    • Data up-to-dateness through synchronization with the main database.

    LUNA Vinder Module is provided as a separate distribution containing an administrator's guide with information about the module's operation, installation instructions (manual and with Docker Compose), OpenAPI specifications, and startup scripts.

    For more information, see the LUNA Vinder Module administrator's guide.