yangzvv

yangzvv / XET-V2X

Public

XET-V2X is a multimodal fused end-to-end 3D spatiotemporal perception framework for V2X collaboration. It unifies multiview multimodal sensing within a shared spatiotemporal representation, enabling robust detection and tracking under occlusions, limited viewpoints, and communication delays in cooperative driving scenarios.

19
2
100% credibility
Found Apr 23, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

XET-V2X is a multimodal fused end-to-end 3D spatiotemporal perception framework for vehicle-to-everything (V2X) collaboration, supporting multiview image-LiDAR fusion, detection, tracking, and compatibility with specific V2X datasets.

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is XET-V2X?

XET-V2X is a Python-based end-to-end framework for 3D spatiotemporal perception in V2X collaboration, fusing multiview multimodal data like camera and LiDAR into a shared representation for robust detection and tracking. It tackles occlusions, limited viewpoints, and communication delays in cooperative driving scenarios, delivering outputs like 3D bounding boxes and trajectories across datasets such as V2X-Seq-SPD and V2X-Sim. Developers get pre-trained checkpoints for quick evaluation on single-agent or multi-agent setups.

Why is it gaining traction?

It stands out with communication-delay-aware benchmarks showing fused models beating camera- or LiDAR-only baselines by 20-50% mAP under delays up to 400ms, enabling real-world V2X testing without custom pipelines. Ready-to-run scripts for training, distributed eval, and data conversion lower the barrier for multimodal perception experiments. The extensible design supports V2V and V2I collaboration out of the box.

Who should use this?

Autonomous driving researchers prototyping cooperative perception systems. V2X engineers simulating delays in vehicle-infrastructure setups. Perception devs at startups needing fused camera-LiDAR tracking for occluded urban driving.

Verdict

Grab the checkpoints for benchmarks if you're in V2X research—performance holds up well despite low maturity (19 stars, 1.0% credibility). Docs and Apache 2.0 license make it forkable, but expect tweaks for production.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.