Repository navigation
Add frame_id to ModelDetections, SotState and LaserDetection - #322
Merged
Merged
Conversation
The id of the camera frame a computer vision result was computed from, as assigned once per frame by the X3 Ultra video pipeline. Recorded video carries the same id in a user_data_unregistered SEI message per access unit, so detections can be joined with the exact frame of a recording instead of matched by time. ModelDetections documents the SEI format; 0 means unknown, and the id stays below 2^53 so it is exact as a JavaScript number. Part of BluEye-Robotics/BlueyeCV#33. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
follesoe
approved these changes
Oct 28, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds
uint64 frame_idtoModelDetections(6),SotState(10) andLaserDetection(11): the id of the camera frame a computer vision result was computed from. The X3 Ultra video pipeline assigns it once per frame and also writes it into every recorded frame. Detections can then be joined with the exact frame of a recording, instead of being matched by time. The design is in BluEye-Robotics/BlueyeCV#33.ModelDetectionsdocuments the SEI format the recordings use, once, for all three fields:user_data_unregistered(payload type 5), one per access unit, before the first slice.PREFIX_SEI_NUT(39) in H.265.6b1e2f0c-935d-4a8e-b17a-52d40e3c91f6, thenframe_idas uint64 BE, then the CLOCK_MONOTONIC capture time in ns as uint64 BE.0means unknown. The id stays below 2^53, so the TypeScript clients decode it exactly.Testing
protolint.shis clean.cross-jetson-compile), Rustcargo test, the TypeScript generate step,dotnet pack, and the Kotlin build and smoke test all pass.frame_iddecoded fromObjectDetectionsTelon node_comm's telemetry socket matched recorded frames for all three detection models.Related
Design and end-to-end results: BluEye-Robotics/BlueyeCV#33.
frame_idin blueye.protocol and the SEI formatframe_idin the ROS messagesframe_idin becv resultsMerge order: ProtocolDefinitions and p2_msgs first, then BlueyeCV and gst_rtsp_record, then p2_drone. Deploy all five together.
🤖 Generated with Claude Code