FindIt - Multi-Label Object Detection: leaderboard

Metric: Average F1@0.5 (%) over multi-label object detection (every instance of several categories, labelled) on 1,000-query subsets of Pascal VOC, OpenImages and iGround; each model scored with the box representation, text or JSON output and JSON key that a two-stage probe on Pascal VOC found best for it; boxes Hungarian-matched to the ground truth, a match counts at IoU 0.5 or more (with the right label for multi-label queries); unparseable outputs count as empty; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 10 models tracked.

Top models

#ModelScore
1Qwen 3 VL 8B Instruct63.8
2Qwen 3.5 9B (Non-reasoning)60.9
3Gemini 2.5 Flash (Non-reasoning)60.3
4Gemma 4 E4B58
5InternVL3-8B57.2
6Qwen 3.5 9B (Thinking)54
7GPT-5.4 (Non-reasoning)44.5
8Qwen 2.5 VL 7B Instruct32.7
9Claude Sonnet 4.527

Interactive version: theaggregate.ai/benchmark?slug=findit-multi-label-object-detection · How It Works · Data refreshed daily, snapshot 2026-09-29.