Introduction: Interested in "Object Tracking (ByteTrack)" (GSoC 2026)

[Raakshass]

Hi Dariusz Trawinski and Damian Kalinowski,

I'm SIDDHANT JAIN, a pre-final year student at IIT Roorkee. I'm interested in the "Object tracking in MP with OpenVINO inference" project for GSoC 2026.

Background:

• Completed a prerequisite PR: keras-team/keras#22028 (Implemented numpy.logaddexp2 for OpenVINO backend, demonstrating familiarity with the OpenVINO contribution workflow).
• Experience with Python and C++ fundamentals.
• Currently studying the ByteTrack algorithm and MediaPipe architecture.

Questions:

1. Are there any specific YOLO model versions recommended for this project (e.g., YOLOv8, YOLOv5)?
2. Should the MediaPipe graph be designed for single-stream or multi-stream input?
3. Are there any existing OpenVINO + MediaPipe examples I should study for ramp-up?

Looking forward to your guidance!

[adrianboguszewski]

Hi there, OpenVINO has not yet been officially announced as a 2026 organization. We don't know if we're accepted. Please wait for the official announcement.

[Raakshass]

Hi @adrianboguszewski,
Thank you for the clarification! I understand that the official announcement is pending.
In the meantime, I'll continue contributing to OpenVINO to build my familiarity with the codebase. I've already submitted a PR for aten::max_unpool2d (#33709) and have a merged prerequisite contribution for the Keras OpenVINO backend.
I'll keep an eye on the official GSoC 2026 announcement. If OpenVINO gets accepted, I'd love to discuss the ByteTrack project further with the mentors.
Thanks for the heads up!