AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

CCCaption: Dual-Reward Reinforcement Learning for Complete and Correct Image Captioning

Zhijiang Tang, Linhua Wang, Jiaxin Qi, Weihao Jiang, Peng Hou, Anxiang Zeng, Jianqiang Huang

CVPR 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Reinforcement learning for visual reasoning · 7 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

CCCaption trains captioners with GRPO using completeness and correctness rewards.

Problem

Captioning references are incomplete or incorrect.

Approach

Dual reward combining completeness and correctness under GRPO.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add RL training with dual rewards to Qwen3-VL-2BQwen3-VL-2B · CapArena (avg)win-minus-loss score: 41.39 -> 46.39Improved
Add RL training with dual rewards to Qwen3-VL-2BQwen3-VL-2B · Hallucinations evaluation (avg)correctness score: 70.19 -> 70.57Improved
Use RL with completeness reward only (no correctness reward, as in CapRL)Qwen2.5-VL-3B · Hallucinations evaluation (avg)correctness score: 67.77 -> 66.91Hurt
Add correctness reward to completeness reward (ablation)CCCaption-2B · ChartQA/OCR Prism and Hallucination avgaverage score: 67.56 -> 70.07Improved
Add completeness reward to correctness reward (ablation)CCCaption-2B · ChartQA/OCR Prism and Hallucination avgaverage score: 65.93 -> 70.07Improved
Add dynamic query sampling (ablation)CCCaption-2B · ChartQA/OCR Prism and Hallucination avgaverage score: 69.15 -> 70.07Improved
Use correctness-only reward instead of completeness-only on OCR (ablation)CCCaption-2B · OCRBench Prismaccuracy: 50.16 -> 40.45Hurt
Use multi-MLLM CCaption-44k queries instead of single-MLLM Vanilla queries (ablation)Qwen3-VL-2B · Prism evaluation (4-benchmark avg)average accuracy: 44.07 -> 45.94Improved
Scale CCaption training data from 8k to 44k samples (ablation)Qwen3-VL-2B · Prism evaluation (4-benchmark avg)average accuracy: 44.30 -> 45.94Improved
Scale CCaption training data from 16k to 44k samples (ablation)Qwen3-VL-2B · Prism evaluation (4-benchmark avg)average accuracy: 44.94 -> 45.94Improved
Train on CCaption-44k with RL instead of no RL (ablation)Qwen3-VL-2B · Prism evaluation (4-benchmark avg)average accuracy: 43.25 -> 45.94Improved
Add CCCaption training to 2B model vs CapRL-3BCCCaption-2B · Prism evaluation (avg)accuracy: 51.07 -> 52.80Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy