WildRoadBench - Pavement Debris: leaderboard

Metric: F1 (%) of predicted boxes at IoU 0.5 for the pavement debris scene class (85 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: Not forecast. 25 models tracked.

Top models

#ModelScore
1Gemini 3 Pro57.3
2Claude Sonnet 4.641.4
3Qwen 3.6 Plus41.2
4Qwen 2.5 VL 7B Instruct40.6
5Gemini 3.1 Flash Lite32.8
6Qwen 3 VL 8B Instruct27.4
7Qwen 3 VL 8B (Thinking)18.6
8Kimi K2.511
9InternVL3-8B11
10Gemini 2.5 Flash9.5
11GPT-5.4 (2026-03-05)9.1
12InternVL2-8B4.4
13InternVL3-14B2.6
14Claude Opus 4.60.8
15GPT-4o (2024-05-13)0

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