WildRoadBench - Guardrail Damage: leaderboard
The guardrail damage class of WildRoadBench (2026), an aerial road-damage grounding benchmark: models box every instance in wild drone images, 20 ground-truth boxes in all. The score is F1 at IoU 0.5 in percent; higher is better.
Metric: F1 (%) of predicted boxes at IoU 0.5 for the guardrail damage scene class (20 ground-truth boxes), VLM Track, zero-shot box grounding on the full 1,061-image aerial set (1,394 ground-truth boxes); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 25 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 38.1 | #145 |
| 2 | Claude Sonnet 4.6 | 37.8 | #173 |
| 3 | Qwen3-VL-32B-Instruct-AWQ | 34.3 | |
| 4 | Qwen 3.6 Plus | 27 | #238 |
| 5 | Kimi K2.5 | 25 | #266 |
| 6 | Gemini 3.1 Flash Lite | 23.8 | #371 |
| 7 | Qwen 3 VL 8B Instruct | 17.6 | #1029 |
| 8 | Qwen3-VL-32B-Thinking-AWQ | 17 | |
| 9 | InternVL3-8B | 16.7 | #1782 |
| 10 | Qwen 3.5 VL 9B | 15.4 | |
| 11 | Qwen 3.5 VL 4B | 11.4 | |
| 12 | Gemini 2.5 Flash | 11.1 | #463 |
| 13 | GPT-5.4 (2026-03-05) | 10 | #92 |
| 14 | Qwen2.5-VL-32B-Instruct-AWQ | 5.7 | |
| 15 | Qwen 3 VL 8B (Thinking) | 5.4 | #910 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=wildroadbench-guardrail-damage · How It Works · Data refreshed daily, snapshot 2026-11-21.