GRAZE: Grounded Refinement and Motion-Aware Zero-Shot Event Localization

2026-04-01

AI Summary

Accurately localizing the first point of contact (FPOC) between players and dummies in untrimmed American football training videos is highly challenging due to camera motion, scene clutter, visual similarity among actors, and rapid pose changes. This work proposes a zero-shot, training-free method that decouples interaction candidate discovery from pixel-level contact verification for the first time. It leverages Grounding DINO to generate interaction candidates, employs motion-aware temporal reasoning to filter potential contact frames, and introduces SAM2 for explicit pixel-level contact confirmation. The approach operates without reliance on detection confidence scores or annotated data, achieving successful FPOC estimation in 97.4% of 738 videos, with 77.5% of predictions within ±10 frames and 82.7% within ±20 frames of the ground truth, substantially improving both robustness and precision.

Abstract

American football practice generates video at scale, yet the interaction of interest occupies only a brief window of each long, untrimmed clip. Reliable biomechanical analysis, therefore, depends on spatiotemporal localization that identifies both the interacting entities and the onset of contact. We study First Point of Contact (FPOC), defined as the first frame in which a player physically touches a tackle dummy, in unconstrained practice footage with camera motion, clutter, multiple similarly equipped athletes, and rapid pose changes around impact. We present GRAZE, a training-free pipeline for FPOC localization that requires no labeled tackle-contact examples. GRAZE uses Grounding DINO to discover candidate player-dummy interactions, refines them with motion-aware temporal reasoning, and uses SAM2 as an explicit pixel-level verifier of contact rather than relying on detection confidence alone. This separation between candidate discovery and contact confirmation makes the approach robust to cluttered scenes and unstable grounding near impact. On 738 tackle-practice videos, GRAZE produces valid outputs for 97.4% of clips and localizes FPOC within $\pm$ 10 frames on 77.5% of all clips and within $\pm$ 20 frames on 82.7% of all clips. These results show that frame-accurate contact onset localization in real-world practice footage is feasible without task-specific training.

Problem

• event localization
• first point of contact
• zero-shot
• video analysis
• spatiotemporal localization

Innovation

• zero-shot event localization
• motion-aware temporal reasoning
• pixel-level contact verification
• training-free pipeline
• grounded refinement