AI Paper Atlas

Research map / Benchmarks / RefCOCOg

RefCOCOg leaderboard

82 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on RefCOCOg. Best accuracy: 91.7 (Florence-2-L, CVPR 2024).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

RefCOCOg · accuracy

26 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
191.7Florence-2-LFlorence-2: Advancing a Unified Representation for a Variety of Vision Tasks
CVPR 2024 · Table 5
CLIP and zero-shot
290.8MATA (Domain-Specific)
InternVL2.5 (8B) + Qwen3 (4B) controller
MATA: A Trainable Hierarchical Automaton System for Multi-Agent Visual Reasoning
ICLR 2026 · Table 3
LLM agents
390.79CogVLM-GroundingCogVLM: Visual Expert for Pretrained Language Models
NeurIPS 2024 · Grounding table
Human and embodied modeling
490.45CogCoM
CogVLM
CogCoM: A Visual Language Model with Chain-of-Manipulations Reasoning
ICLR 2025 · Table: Results on VG benchmark
Multimodal LLMs
588.04AerialVG (Ours)
Swin-L
AerialVG: A Challenging Benchmark for Aerial Visual Grounding by Exploring Positional Relations
ICCV 2025 · Table 2
Human and embodied modeling
687.01Groma
Vicuna-7B v1.5
Groma: Localized Visual Tokenization for Grounding Multimodal Large Language Models
ECCV 2024 · Table 2
Multimodal LLMs
786.19Octopus
Vicuna-7B-v1.5
Octopus: A Multi-modal LLM with Parallel Recognition and Sequential Understanding
NeurIPS 2024 · Table 1
Multimodal LLMs
885.32PromViL‡
MiniGPTv2 7B
Progressive Multi-granular Alignments for Grounded Reasoning in Large Vision-Language Models
AAAI 2025 · Table 2
Multimodal LLMs
985.1IAA-14
Llama3-8B
IAA: Inner-Adaptor Architecture Empowers Frozen Large Language Model with Multimodal Capabilities
AAAI 2025 · Table: Comparisons on visual g
Multimodal LLMs
1083.52DiffusionREC w/ DDIMDiffusionREC: Diffusion Model with Adaptive Condition for Referring Expression Comprehension
AAAI 2025 · Table 1
Autoregressive and latent generation
16 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

RefCOCOg · accuracy (val)

25 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
192.8Phi3V-dist
Phi-3.5-vision
Teaching Metric Distance to Discrete Autoregressive Language Models
ICLR 2026 · Table 1
Language, safety and interpretability
291.2Florence-2-LFlorence-2: Advancing a Unified Representation for a Variety of Vision Tasks
CVPR 2024 · Table 5
CLIP and zero-shot
389.32CogCoM
CogVLM
CogCoM: A Visual Language Model with Chain-of-Manipulations Reasoning
ICLR 2025 · Table: Results on VG benchmark
Multimodal LLMs
489CofactGVRCofactGVR: Counterfactual Intervention for Grounded Visual Reasoning
ICML 2026 · Table 3
Multimodal LLMs
588.43ForeSight (7B)
Qwen2.5-VL-7B
See Further, Think Deeper: Advancing VLM's Reasoning Ability with Low-level Visual Cues and Reflection
CVPR 2026 · Table 6
Multimodal LLMs
686.52AerialVG (Ours)
Swin-L
AerialVG: A Challenging Benchmark for Aerial Visual Grounding by Exploring Positional Relations
ICCV 2025 · Table 2
Human and embodied modeling
785IAA-14
Llama3-8B
IAA: Inner-Adaptor Architecture Empowers Frozen Large Language Model with Multimodal Capabilities
AAAI 2025 · Table: Comparisons on visual g
Multimodal LLMs
884.25Octopus
Vicuna-7B-v1.5
Octopus: A Multi-modal LLM with Parallel Recognition and Sequential Understanding
NeurIPS 2024 · Table 1
Multimodal LLMs
983.94DiffusionRECDiffusionREC: Diffusion Model with Adaptive Condition for Referring Expression Comprehension
AAAI 2025 · Table 1
Autoregressive and latent generation
1083.35MDETR
ENB3
MDETR - Modulated Detection for End-to-End Multi-Modal Understanding
ICCV 2021 · Table 2
Human and embodied modeling
15 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

RefCOCOg · cIoU

20 papers, each with its best reported number for its own method; higher is better.

10 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

RefCOCOg · cIoU (val)

16 papers, each with its best reported number for its own method; higher is better.

6 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

RefCOCOg · CIDEr

13 papers, each with its best reported number for its own method; higher is better.

#CIDEr ↑MethodPaperTopic
1118.4ClawMachineX-7B
LaVIT-2-7B
ClawMachine: Learning to Fetch Visual Tokens for Referential Comprehension
ICLR 2025 · Table: visual referring result
Multimodal LLMs
2118.2FineCaption-8B
LLaMA-3.1-8B-Instruct
FINECAPTION: Compositional Image Captioning Focusing on Wherever You Want at Any Granularity
CVPR 2025 · Table 6
Multimodal LLMs
3115.7DynRefer
EVA, Q-former, FlanT5XL
DynRefer: Delving into Region-level Multimodal Tasks via Dynamic Resolution
CVPR 2025 · Table 1
Human and embodied modeling
4113.8ROD-MLLM
Vicuna-7B v1.5
ROD-MLLM: Towards More Reliable Object Detection in Multimodal Large Language Models
CVPR 2025 · Table 5
Open-set and few-shot detection
5109.9RGPT
CLIP ViT-L + Vicuna-7B
RegionGPT: Towards Region Understanding Vision Language Model
CVPR 2024 · Table: Performance on the regi
Multimodal LLMs
6109.7Omni-RGPT
Llama-2 7B
Omni-RGPT: Unifying Image and Video Region-level Understanding via Token Marks
CVPR 2025 · Table 5
Video-language understanding
7109.2Alpha-CLIP+LLaVA
LLaVA-1.5
Alpha-CLIP: A CLIP Model Focusing on Wherever You Want
CVPR 2024 · Table 6
CLIP and zero-shot
8106GLaMM (FT)GLaMM: Pixel Grounding Large Multimodal Model
CVPR 2024 · Table 5
Human and embodied modeling
9105.9ViP-LLaVA-Base-7B
Vicuna v1.5-7B
ViP-LLaVA: Making Large Multimodal Models Understand Arbitrary Visual Prompts
CVPR 2024 · Region captioning table
Multimodal LLMs
10103ASM-FT
Husky-7B
The All-Seeing Project: Towards Panoptic Visual Recognition and Understanding of the Open World
ICLR 2024 · Table: Performance on the regi
Visual question answering
3 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

RefCOCOg · METEOR

12 papers, each with its best reported number for its own method; higher is better.

#METEOR ↑MethodPaperTopic
120.8ASM-FT
Husky-7B
The All-Seeing Project: Towards Panoptic Visual Recognition and Understanding of the Open World
ICLR 2024 · Table: Performance on the regi
Visual question answering
220.7LocPred + NegPred + RevLoc
LocVLM-B
Learning to Localize Objects Improves Spatial Reasoning in Visual-LLMs
CVPR 2024 · Ablations table
Multimodal LLMs
318.1DynRefer
EVA, Q-former, FlanT5XL
DynRefer: Delving into Region-level Multimodal Tasks via Dynamic Resolution
CVPR 2025 · Table 1
Human and embodied modeling
417.5FineCaption-8B
LLaMA-3.1-8B-Instruct
FINECAPTION: Compositional Image Captioning Focusing on Wherever You Want at Any Granularity
CVPR 2025 · Table 6
Multimodal LLMs
517.4ClawMachineX-7B
LaVIT-2-7B
ClawMachine: Learning to Fetch Visual Tokens for Referential Comprehension
ICLR 2025 · Table: visual referring result
Multimodal LLMs
617Omni-RGPT
Llama-2 7B
Omni-RGPT: Unifying Image and Video Region-level Understanding via Token Marks
CVPR 2025 · Table 5
Video-language understanding
716.9RGPT
CLIP ViT-L + Vicuna-7B
RegionGPT: Towards Region Understanding Vision Language Model
CVPR 2024 · Table: Performance on the regi
Multimodal LLMs
816.7Alpha-CLIP+LLaVA
LLaVA-1.5
Alpha-CLIP: A CLIP Model Focusing on Wherever You Want
CVPR 2024 · Table 6
CLIP and zero-shot
916.6ChatterBox
LLaVA-13B
ChatterBox: Multimodal Referring and Grounding with Chain-of-Questions
AAAI 2025 · Table 3
Multimodal LLMs
1016.6ViP-LLaVA-Base-7B
Vicuna v1.5-7B
ViP-LLaVA: Making Large Multimodal Models Understand Arbitrary Visual Prompts
CVPR 2024 · Region captioning table
Multimodal LLMs
2 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.