WildRoadBench - Road Crack: leaderboard

Metric: F1 (%) of predicted boxes at IoU 0.5 for the road crack scene class (561 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 March 2028. 25 models tracked.

Top models

#ModelScore
1Gemini 3 Pro33.3
2Qwen 3.6 Plus25.4
3Gemini 3.1 Flash Lite16.2
4Qwen 2.5 VL 7B Instruct15.3
5Claude Sonnet 4.611.3
6GPT-5.4 (2026-03-05)10.8
7Qwen 3 VL 8B Instruct6.2
8Kimi K2.54
9Claude Opus 4.62.1
10Gemini 2.5 Flash1.9
11Qwen 3 VL 8B (Thinking)1.9
12InternVL3-8B0.8
13GPT-4o (2024-05-13)0.7
14InternVL3-14B0.6
15InternVL2-8B0

Interactive version: theaggregate.ai/benchmark?slug=wildroadbench-road-crack · How It Works · Data refreshed daily, snapshot 2026-10-07.