# Polos: Multimodal Metric Learning from Human Feedback for Image Captioning

arXiv: https://arxiv.org/abs/2402.18091 · HTML: https://mlsanity.com/paper/2402.18091/

## Key facts

- Cited by 46 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 55 of its 83 references that are in the map.
- First posted on arXiv 28 Feb 2024.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Diffusion and multimodal model alignment](https://mlsanity.com/alignment/diffusion-and-multimodal-model-alignment/)

## Builds on (most cited first, 30 of 55)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://mlsanity.com/paper/2201.12086/) (2022)
- [Unsupervised Cross-lingual Representation Learning at Scale](https://mlsanity.com/paper/1911.02116/) (2019)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [Scaling Instruction-Finetuned Language Models](https://mlsanity.com/paper/2210.11416/) (2022)
- [CLIPScore: A Reference-free Evaluation Metric for Image Captioning](https://mlsanity.com/paper/2104.08718/) (2021)
- [SimCSE: Simple Contrastive Learning of Sentence Embeddings](https://mlsanity.com/paper/2104.08821/) (2021)
- [CIDEr: Consensus-based Image Description Evaluation](https://mlsanity.com/paper/1411.5726/) (2014)
- [Show, Attend and Tell: Neural Image Caption Generation with Visual Attention](https://mlsanity.com/paper/1502.03044/) (2015)
- [Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering](https://mlsanity.com/paper/1707.07998/) (2017)
- [BLEURT: Learning Robust Metrics for Text Generation](https://mlsanity.com/paper/2004.04696/) (2020)
- [COMET: A Neural Framework for MT Evaluation](https://mlsanity.com/paper/2009.09025/) (2020)
- [SemEval-2017 Task 1: Semantic Textual Similarity - Multilingual and Cross-lingual Focused Evaluation](https://mlsanity.com/paper/1708.00055/) (2017)
- [Open-vocabulary Object Detection via Vision and Language Knowledge Distillation](https://mlsanity.com/paper/2104.13921/) (2021)
- [SPICE: Semantic Propositional Image Caption Evaluation](https://mlsanity.com/paper/1607.08822/) (2016)
- [OFA: Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework](https://mlsanity.com/paper/2202.03052/) (2022)
- [BARTScore: Evaluating Generated Text as Text Generation](https://mlsanity.com/paper/2106.11520/) (2021)
- [RegionCLIP: Region-based Language-Image Pretraining](https://mlsanity.com/paper/2112.09106/) (2021)
- [Self-critical Sequence Training for Image Captioning](https://mlsanity.com/paper/1612.00563/) (2016)
- [GIT: A Generative Image-to-text Transformer for Vision and Language](https://mlsanity.com/paper/2205.14100/) (2022)
- [nocaps: novel object captioning at scale](https://mlsanity.com/paper/1812.08658/) (2018)
- [MoverScore: Text Generation Evaluating with Contextualized Embeddings and Earth Mover Distance](https://mlsanity.com/paper/1909.02622/) (2019)
- [TextCaps: a Dataset for Image Captioning with Reading Comprehension](https://mlsanity.com/paper/2003.12462/) (2020)

## Built on by (most cited first, 30 of 46)

- [DreamReward: Text-to-3D Generation with Human Preference](https://mlsanity.com/paper/2403.14613/) (2024)
- [FLEUR: An Explainable Reference-Free Evaluation Metric for Image Captioning Using a Large Multimodal Model](https://mlsanity.com/paper/2406.06004/) (2024)
- [CapArena: Benchmarking and Analyzing Detailed Image Captioning in the LLM Era](https://mlsanity.com/paper/2503.12329/) (2025)
- [SAM-R1: Leveraging SAM for Reward Feedback in Multimodal Segmentation via Reinforcement Learning](https://mlsanity.com/paper/2505.22596/) (2025)
- [G-VEval: A Versatile Metric for Evaluating Image and Video Captions Using GPT-4o](https://mlsanity.com/paper/2412.13647/) (2024)
- [Image Captioning Evaluation in the Age of Multimodal LLMs: Challenges and Future Perspectives](https://mlsanity.com/paper/2503.14604/) (2025)
- [Positive-Augmented Contrastive Learning for Vision-and-Language Evaluation and Training](https://mlsanity.com/paper/2410.07336/) (2024)
- [Patch Matters: Training-free Fine-grained Image Caption Enhancement via Local Perception](https://mlsanity.com/paper/2504.06666/) (2025)
- [UniLumos: Fast and Unified Image and Video Relighting with Physics-Plausible Feedback](https://mlsanity.com/paper/2511.01678/) (2025)
- [DENEB: A Hallucination-Robust Automatic Evaluation Metric for Image Captioning](https://mlsanity.com/paper/2409.19255/) (2024)
- [YesBut: A High-Quality Annotated Multimodal Dataset for evaluating Satire Comprehension capability of Vision-Language Models](https://mlsanity.com/paper/2409.13592/) (2024)
- [VLM Judges Can Rank but Cannot Score: Task-Dependent Uncertainty in Multimodal Evaluation](https://mlsanity.com/paper/2604.25235/) (2026)
- [JEEM: Vision-Language Understanding in Four Arabic Dialects](https://mlsanity.com/paper/2503.21910/) (2025)
- [EXPERT: An Explainable Image Captioning Evaluation Metric with Structured Explanations](https://mlsanity.com/paper/2506.24016/) (2025)
- [Evaluation of Multilingual Image Captioning: How far can we get with CLIP models?](https://mlsanity.com/paper/2502.06600/) (2025)
- [ZINA: Multimodal Fine-grained Hallucination Detection and Editing](https://mlsanity.com/paper/2506.13130/) (2025)
- [VELA: An LLM-Hybrid-as-a-Judge Approach for Evaluating Long Image Captions](https://mlsanity.com/paper/2509.25818/) (2025)
- [HEDGE: Hallucination Estimation via Dense Geometric Entropy for VQA with Vision-Language Models](https://mlsanity.com/paper/2511.12693/) (2025)
- [CUS-QA: Local-Knowledge-Oriented Open-Ended Question Answering Dataset](https://mlsanity.com/paper/2507.22752/) (2025)
- [VC-Inspector: Advancing Reference-free Evaluation of Video Captions with Factual Analysis](https://mlsanity.com/paper/2509.16538/) (2025)
- [Multi-modal, Multi-task, Multi-criteria Automatic Evaluation with Vision Language Models](https://mlsanity.com/paper/2412.14613/) (2024)
- [Sightation Counts: Leveraging Sighted User Feedback in Building a BLV-aligned Dataset of Diagram Descriptions](https://mlsanity.com/paper/2503.13369/) (2025)
- [OSMDA: OpenStreetMap-based Domain Adaptation for Remote Sensing VLMs](https://mlsanity.com/paper/2603.11804/) (2026)
- [The Lipreading Gap: Do VSR Models Perceive Visual Speech Like Human Lipreaders?](https://mlsanity.com/paper/2606.07435/) (2026)
- [Towards Unified Benchmark and Models for Multi-Modal Perceptual Metrics](https://mlsanity.com/paper/2412.10594/) (2024)
- [Personalized Image Descriptions from Attention Sequences](https://mlsanity.com/paper/2512.06662/) (2025)
- [DISCODE: Distribution-Aware Score Decoder for Robust Automatic Evaluation of Image Captioning](https://mlsanity.com/paper/2512.14420/) (2025)
- [LLM-Free Image Captioning Evaluation in Reference-Flexible Settings](https://mlsanity.com/paper/2512.21582/) (2025)
- [Surveying the Landscape of Image Captioning Evaluation: A Comprehensive Taxonomy, Trends and Metrics Analysis](https://mlsanity.com/paper/2408.04909/) (2024)
- [DiscoSG: Towards Discourse-Level Text Scene Graph Parsing through Iterative Graph Refinement](https://mlsanity.com/paper/2506.15583/) (2025)

## Abstract (arXiv)

Establishing an automatic evaluation metric that closely aligns with human judgments is essential for effectively developing image captioning models. Recent data-driven metrics have demonstrated a stronger correlation with human judgments than classic metrics such as CIDEr; however they lack sufficient capabilities to handle hallucinations and generalize across diverse images and texts partially because they compute scalar similarities merely using embeddings learned from tasks unrelated to image captioning evaluation. In this study, we propose Polos, a supervised automatic evaluation metric for image captioning models. Polos computes scores from multimodal inputs, using a parallel feature extraction mechanism that leverages embeddings trained through large-scale contrastive learning. To train Polos, we introduce Multimodal Metric Learning from Human Feedback (M2LHF), a framework for developing metrics based on human feedback. We constructed the Polaris dataset, which comprises 131K human judgments from 550 evaluators, which is approximately ten times larger than standard datasets. Our approach achieved state-of-the-art performance on Composite, Flickr8K-Expert, Flickr8K-CF, PASCAL-50S, FOIL, and the Polaris dataset, thereby demonstrating its effectiveness and robustness.