AI Paper Atlas

Research map / Benchmarks / Flickr30K

Flickr30K leaderboard

69 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on Flickr30K. Best CIDEr: 104.6 (MGLMM, AAAI 2025).

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.

Flickr30K · CIDEr

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

#CIDEr ↑MethodPaperTopic
1104.6MGLMM
Vicuna-7B
Instruction-guided Multi-Granularity Segmentation and Captioning with Large Multimodal Model
AAAI 2025 · Table: comprehensive compariso
Human and embodied modeling
295.3GLaMMGLaMM: Pixel Grounding Large Multimodal Model
CVPR 2024 · Table 6
Human and embodied modeling
389.6video-SALMONN 13B
Vicuna-v1.5 13B
video-SALMONN: Speech-Enhanced Audio-Visual Large Language Models
ICML 2024 · SAVE single-modal results tabl
Video-language understanding
489.2CW-DPO (Ours)
Qwen2.5-VL-72B
Dynamics-Aware Preference Optimization for Vision-Language Models
CVPR 2026 · Table 2
Multimodal LLMs
588.5DDT-LLaMA
LLaMA3-8B
Generative Multimodal Pretraining with Discrete Diffusion Timestep Tokens
CVPR 2025 · Table 3
Multimodal LLMs
688.3AKGMA
Qwen2.5-7B
Knowledge-Enhanced Image Captioning with Adaptive Graph-based Multimodal Alignment and LLM
AAAI 2026 · Table 2
Human and embodied modeling
788.2WWT(RL)What and Where the Themes Dominate in Image
AAAI 2019 · Table 2
Open-set and few-shot detection
888OURO-7B
Qwen2-VL-7B-Instruct
OURO: A Self-Bootstrapped Framework for Enhancing Multimodal Scene Understanding
ICCV 2025 · Table 4
Multimodal LLMs
987.7ASM-FT
Husky-7B
The All-Seeing Project: Towards Panoptic Visual Recognition and Understanding of the Open World
ICLR 2024 · Table: Zero-shot performance o
Visual question answering
1087.1BLIVA (Vicuna-7B)
InstructBLIP (Vicuna-7B)
BLIVA: A Simple Multimodal LLM for Better Handling of Text-Rich Visual Questions
AAAI 2024 · Table 2
Visual question answering
28 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

Flickr30K · R@1

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

#R@1 ↑MethodPaperTopic
189.61InternVL†_IB
ImageBind
FreeBind: Free Lunch in Unified Multimodal Space via Knowledge Fusion
ICML 2024 · Results of zero-shot cross-mod
Cross-modal learning
289.13OmniBind-FullOmniBind: Large-scale Omni Multimodal Representation via Binding Spaces
ICLR 2025 · Cross-modal retrieval results
Cross-modal learning
386.1EVA-CLIP+VISA
EVA-CLIP
Visual Abstraction: A Plug-and-Play Approach for Text-Visual Retrieval
ICML 2025 · Table 1
Human and embodied modeling
485.2Florence-2-LFlorence-2: Advancing a Unified Representation for a Variety of Vision Tasks
CVPR 2024 · Table 5
CLIP and zero-shot
579.3Ours
CLIP ViT-B/32
Continual Vision-Language Retrieval via Dynamic Knowledge Rectification
AAAI 2024 · Table 1
Continual fine-tuning of LLMs
677.6dino.txt
DINOv2 ViT-L/14
DINOv2 Meets Text: A Unified Framework for Image- and Pixel-Level Vision-Language Alignment
CVPR 2025 · Table 5
CLIP and zero-shot
772.2DataComp + RLFT
DataComp
Aligning Vision Models with Human Aesthetics in Retrieval: Benchmarks and Algorithms
NeurIPS 2024 · Table: Performance on traditio
Alignment and preferences
868.4TransResNet
ResNeXt-IG-3.5B; Transformer with word pretraining
Engaging Image Captioning via Personality
CVPR 2019 · Retrieval model results table
Human and embodied modeling
964.19GroundVLP w/ VinVL
VinVL-Large
GroundVLP: Harnessing Zero-Shot Visual Grounding from Vision-Language Pre-training and Open-Vocabulary Object Detection
AAAI 2024 · Flickr30k results table
CLIP and zero-shot
1059.53Diff-Prompt
GLIP-T(A)
Diff-Prompt: Diffusion-Driven Prompt Generator with Mask Supervision
ICLR 2025 · Table 1
Text-to-image diffusion
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

Flickr30K · RSUM

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

#RSUM ↑MethodPaperTopic
1547MP-ISMoE (SHERL)
VSE∞
MP-ISMoE: Mixed-Precision Interactive Side Mixture-of-Experts for Efficient Transfer Learning
AAAI 2026 · Table 1
Pruning and efficient tuning
2542.9APSE-IPIK
CLIP ViT-B/32
Adaptive Prompt-Based Semantic Embedding with Inspire Potential of Implicit Knowledge for Cross-Modal Retrieval
AAAI 2025 · Table 1
Cross-modal learning
3537.4UniPT
VSE∞ (ResNeXt-101 + BERT-base)
UniPT: Universal Parallel Tuning for Transfer Learning with Efficient Parameter and Memory
CVPR 2024 · Table 3
Pruning and efficient tuning
4533.8CHAN + HNR
CHAN, BUTD + BERT
Expanding the Scope of Negatives: Boosting Image-Text Matching with Negatives Distribution Guided Learning
AAAI 2025 · Table 4
Human and embodied modeling
5526.3CLIP ViT-B/32 + CUSA
CLIP ViT-B/32
Cross-Modal and Uni-Modal Soft-Label Alignment for Image-Text Retrieval
AAAI 2024 · Table 5
Cross-modal learning
6520.9CODERCODER: Coupled Diversity-Sensitive Momentum Contrastive Learning for Image-Text Retrieval
ECCV 2022 · Table 1
Cross-modal learning
7518.5CHAN
BUTD + BERT
Fine-Grained Image-Text Matching by Cross-Modal Hard Aligning Network
CVPR 2023 · Table 2
Human and embodied modeling
8515.8CORA (1 Obj-Att, 2 Obj-Obj GAT layers)
Faster R-CNN + Bi-GRU
Composing Object Relations and Attributes for Image-Text Matching
CVPR 2024 · Table 3
Open-set and few-shot detection
9513.2NAAFNegative-Aware Attention Framework for Image-Text Matching
CVPR 2022 · Table 1
Human and embodied modeling
10507.8CMCANShow Your Faith: Cross-Modal Confidence-Aware Network for Image-Text Matching
AAAI 2022 · Table 1
Cross-modal learning
1 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

Flickr30K · Image retrieval R@1

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

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