SOTAVerified

Image Captioning

Image Captioning is the task of describing the content of an image in words. This task lies at the intersection of computer vision and natural language processing. Most image captioning systems use an encoder-decoder framework, where an input image is encoded into an intermediate representation of the information in the image, and then decoded into a descriptive text sequence. The most popular benchmarks are nocaps and COCO, and models are typically evaluated according to a BLEU or CIDER metric.

( Image credit: Reflective Decoding Network for Image Captioning, ICCV'19)

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1–25 of 1878 papers

TitleStatusHype
Language-Guided Contrastive Audio-Visual Masked Autoencoder with Automatically Generated Audio-Visual-Text Triplets from Videos—0
Mask-aware Text-to-Image Retrieval: Referring Expression Segmentation Meets Cross-modal Retrieval—0
HalLoc: Token-level Localization of Hallucinations for Vision Language ModelsCode0
Vision Matters: Simple Visual Perturbations Can Boost Multimodal Math ReasoningCode2
A Novel Lightweight Transformer with Edge-Aware Fusion for Remote Sensing Image Captioning—0
ViCrit: A Verifiable Reinforcement Learning Proxy Task for Visual Perception in VLMsCode1
An Open-Source Software Toolkit & Benchmark Suite for the Evaluation and Adaptation of Multimodal Action Models—0
Better Reasoning with Less Data: Enhancing VLMs Through Unified Modality Scoring—0
Edit Flows: Flow Matching with Edit Operations—0
DiscoVLA: Discrepancy Reduction in Vision, Language, and Alignment for Parameter-Efficient Video-Text RetrievalCode1
Dense Retrievers Can Fail on Simple Queries: Revealing The Granularity Dilemma of EmbeddingsCode0
GTR-CoT: Graph Traversal as Visual Chain of Thought for Molecular Structure RecognitionCode0
Hallucination at a Glance: Controlled Visual Edits and Fine-Grained Multimodal Learning—0
Stepwise Decomposition and Dual-stream Focus: A Novel Approach for Training-free Camouflaged Object SegmentationCode0
SRD: Reinforcement-Learned Semantic Perturbation for Backdoor Defense in VLMs—0
Attention-based transformer models for image captioning across languages: An in-depth survey and evaluation—0
Light as Deception: GPT-driven Natural Relighting Against Vision-Language Pre-training Models—0
Puzzled by Puzzles: When Vision-Language Models Can't Take a HintCode1
CLDTracker: A Comprehensive Language Description for Visual TrackingCode0
Document-Level Text Generation with Minimum Bayes Risk Decoding using Optimal TransportCode0
Beam-Guided Knowledge Replay for Knowledge-Rich Image Captioning using Vision-Language Model—0
Correlating instruction-tuning (in multimodal models) with vision-language processing (in the brain)Code0
SATORI-R1: Incentivizing Multimodal Reasoning with Spatial Grounding and Verifiable RewardsCode1
TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP—0
Scaling Up Biomedical Vision-Language Models: Fine-Tuning, Instruction Tuning, and Multi-Modal LearningCode4
Show:102550
← PrevPage 1 of 76Next →
All datasetsVizWiz 2020 test-devCOCO Captionsnocaps in-domainnocaps near-domainnocaps out-of-domainnocaps entireCOCO (Common Objects in Context)VizWiz 2020 testnocaps-XD entirenocaps-val-in-domainnocaps-val-overallnocaps-XD in-domain

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1IBM Research AICIDEr80.67—Unverified
2CASIA_IVACIDEr79.15—Unverified
3feixiangCIDEr77.31—Unverified
4wocaoCIDEr77.21—Unverified
5lamiwab172CIDEr75.93—Unverified
6RUC_AIM3CIDEr73.52—Unverified
7funasCIDEr73.51—Unverified
8SRC-B_VCLabCIDEr73.47—Unverified
9spartaCIDEr73.41—Unverified
10x-vizCIDEr73.26—Unverified
#ModelMetricClaimedVerifiedStatus
1VALORCIDER152.5—Unverified
2VASTCIDER149—Unverified
3Virtex (ResNet-101)CIDER94—Unverified
4KOSMOS-1 (1.6B) (zero-shot)CIDER84.7—Unverified
5BLIP-FuseCapCLIPScore78.5—Unverified
6mPLUGBLEU-446.5—Unverified
7OFABLEU-444.9—Unverified
8GITBLEU-444.1—Unverified
9BLIP-2 ViT-G OPT 2.7B (zero-shot)BLEU-443.7—Unverified
10BLIP-2 ViT-G OPT 6.7B (zero-shot)BLEU-443.5—Unverified
#ModelMetricClaimedVerifiedStatus
1PaLICIDEr149.1—Unverified
2GIT2, Single ModelCIDEr124.18—Unverified
3GIT, Single ModelCIDEr122.4—Unverified
4PaLICIDEr121.09—Unverified
5CoCa - Google BrainCIDEr117.9—Unverified
6Microsoft Cognitive Services teamCIDEr112.82—Unverified
7Single ModelCIDEr108.98—Unverified
8GRIT (zero-shot, no VL pretraining, no CBS)CIDEr105.9—Unverified
9FudanFVLCIDEr104.9—Unverified
10FudanWYZCIDEr104.25—Unverified
#ModelMetricClaimedVerifiedStatus
1GIT2, Single ModelCIDEr125.51—Unverified
2PaLICIDEr124.35—Unverified
3GIT, Single ModelCIDEr123.92—Unverified
4CoCa - Google BrainCIDEr120.73—Unverified
5Microsoft Cognitive Services teamCIDEr115.54—Unverified
6Single ModelCIDEr110.76—Unverified
7FudanFVLCIDEr109.33—Unverified
8FudanWYZCIDEr108.04—Unverified
9IEDA-LABCIDEr100.15—Unverified
10firetheholeCIDEr99.51—Unverified
#ModelMetricClaimedVerifiedStatus
1PaLICIDEr126.67—Unverified
2GIT2, Single ModelCIDEr122.27—Unverified
3GIT, Single ModelCIDEr122.04—Unverified
4CoCa - Google BrainCIDEr121.69—Unverified
5Microsoft Cognitive Services teamCIDEr110.14—Unverified
6Single ModelCIDEr109.49—Unverified
7FudanFVLCIDEr106.55—Unverified
8FudanWYZCIDEr103.75—Unverified
9HumanCIDEr91.62—Unverified
10firetheholeCIDEr88.54—Unverified