HomeAI & Machine LearningNMS Suppression Graph: 2D Arc Diagram of Greedy Box Filtering

NMS Suppression Graph: 2D Arc Diagram of Greedy Box Filtering

A 2D arc-diagram view of Non-Maximum Suppression: candidate detection boxes become rows sorted by confidence, with overlap arcs showing which boxes the greedy algorithm keeps or suppresses at each IoU threshold.

AI & Machine Learning2DModerate60 FPS📱 Mobile-adapted⇄ 3D version
2d-computer-vision-data-science ↗ Open standalone

The 3D version of this simulation renders a cloud of overlapping detector proposal boxes as extruded bars over a ground plane. This 2D companion keeps the exact same detector model and the exact same greedy Non-Maximum Suppression loop — same IoU formula, same confidence/IoU thresholds, same sort-then-suppress order — but represents it natively in the plane as a sorted arc diagram: one row per candidate box, ranked by confidence, with overlap arcs revealing which boxes suppress which as the greedy loop walks down the list.

⚙ Under the hood

Watch the same greedy Non-Maximum Suppression algorithm run on a sorted arc diagram instead of spatial boxes: each candidate detection becomes a row ranked by confidence, and overlap arcs light up red exactly when the greedy loop suppresses one row because of another.

computer visionobject detectionNMSIoUalgorithmsgraph diagram

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install

What did you find?

Add reproduction steps (optional)