The DOI matches its Crossref record; 19 of the 19 references it cites were found in a registry or an index.
Object Hallucination in Image Captioning
Open accessAbstract
Despite continuously improving performance, contemporary image captioning models are prone to "hallucinating" objects that are not actually in a scene.One problem is that standard metrics only measure similarity to ground truth captions and may not fully capture image relevance.In this work, we propose a new image relevance metric to evaluate current models with veridical visual labels and assess their rate of object hallucination.We analyze how captioning model architectures and learning objectives contribute to object hallucination, explore when hallucination is likely due to image misclassification or language priors, and assess how well current sentence metrics capture object hallucination.We investigate these questions on the standard image captioning benchmark, MSCOCO, using a diverse set of models.Our analysis yields several interesting findings, including that models which score best on standard sentence metrics do not always have lower hallucination and that models which hallucinate more tend to make errors driven by language priors. * Denotes equal contribution.NBT: A woman talking on a cell phone while sitting on a bench.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, it is also apparent that deep learning based generation is prone to hallucinate unintended text, which degrades the system performance and fails to meet user expectations in many real-world scenarios.”
From Survey of Hallucination in Natural Language Generation (2023) · cites this paper as [222]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Large language model (LLM) systems, such as ChatGPT1 or Gemini2, can show impressive reasoning and question-answering capabilities but often ‘hallucinate’ false outputs and unsubstantiated answers3,4.”
From Detecting hallucinations in large language models using semantic entropy (2024) · cites this paper as [4]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“This is because there is another contributor of hallucinations - training and modeling choices of neural models [121, 215, 222, 263].”
From Survey of Hallucination in Natural Language Generation (2023) · cites this paper as [222]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 2×
“All the baseline models employ features extracted from the fourth layer of ResNet-101 (He et al., 2016), except for the GAN model which employs ResNet-152.”
MethodHochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 9×
“We also consider LRCN (Donahue et al., 2015) which considers visual information at each time step, as opposed to just initializing the LSTM hidden state with extracted features.”
BackgroundMethodInfluentialMicrosoft COCO: Common Objects in Context
Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., & Zitnick, C. L. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10602-1_48 · Cited by 45,460
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
Xu, K., Ba, J., Kiros, R., Cho, K., Courville, A., Salakhutdinov, R., Zemel, R., & Bengio, Y. (2015) · arXiv · doi:10.48550/arxiv.1502.03044 · Cited by 8,248 · cited 2×
“For attention based models, we consider Att2In (Rennie et al., 2017), which is similar to the original attention based model proposed by (Xu et al., 2015), except the image feature is only input into the cell gate as this was shown to lead to better performance.”
MethodResultLong-term recurrent convolutional networks for visual recognition and description
Donahue, J., Hendricks, L. A., Guadarrama, S., Rohrbach, M., Venugopalan, S., Darrell, T., & Saenko, K. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298878 · Cited by 5,228 · cited 2×
“We also consider LRCN (Donahue et al., 2015) which considers visual information at each time step, as opposed to just initializing the LSTM hidden state with extracted features.”
MethodSelf-citationVisual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations
Krishna, R., Zhu, Y., Groth, O., Johnson, J., Hata, K., Kravitz, J., Chen, S., Kalantidis, Y., Li, L.-J., Shamma, D. A., et al. (2017) · International Journal of Computer Vision · doi:10.1007/s11263-016-0981-7 · Cited by 5,107 · cited 2×
“his is perhaps because bounding box proposals come from the MSCOCO detection task and are thus “in-domain” as opposed to the TopDown-BB model which relies on proposals learned from the Visual Genome (Krishna et al., 2017) dataset. Second, frequently training models with the self-critical loss actually increases the am…”
MethodCIDEr: Consensus-based image description evaluation
Vedantam, R., Zitnick, C. L., & Parikh, D. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7299087 · Cited by 4,845 · cited 6×
“Likewise, our work is related to other work which aims to build better evaluation tools ((Vedantam et al., 2015), (Anderson et al., 2016), (Cui et al., 2018)).”
BackgroundMethodInfluential(2005) · Cited by 4,152
Self-Critical Sequence Training for Image Captioning
Rennie, S. J., Marcheret, E., Mroueh, Y., Ross, J., & Goel, V. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.131 · Cited by 2,231 · cited 5×
“For non-attention based methods we consider the FC model from Rennie et al. (2017) which incorporates visual information by initializing the LSTM hidden state with high level image features.”
MethodInfluentialSPICE: Semantic Propositional Image Caption Evaluation
Anderson, P., Fernando, B., Johnson, M., & Gould, S. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46454-1_24 · Cited by 1,986 · cited 3×
“Likewise, our work is related to other work which aims to build better evaluation tools ((Vedantam et al., 2015), (Anderson et al., 2016), (Cui et al., 2018)).”
BackgroundMethod
Research works printed but not linked: 3
- Andrej Karpathy and Li Fei-Fei. 2015. Deep visual- semantic alignments for generating image descrip- tions. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3128-3137.
- Mert Kilickaya, Aykut Erdem, Nazli Ikizler-Cinbis, and Erkut Erdem. 2016. Re-evaluating automatic metrics for image captioning. In European Chapter of the Association for Computational Linguistics.
- Anna Rohrbach, Makarand Tapaswi, Atousa Torabi, Tegan Maharaj, Marcus Rohrbach, Sanja Fi- dler Christopher Pal, and Bernt Schiele. 2017a. The Joint Video and Language Understand- ing Workshop: MovieQA and The Large Scale Movie Description Challenge (LSMDC). https://sites.google.com/site/ describingmovies/lsmdc-2017.
Cited by
Sorted by citations
Survey of Hallucination in Natural Language Generation
Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y. J., Madotto, A., & Fung, P. (2023) · ACM Computing Surveys · doi:10.1145/3571730 · Cited by 5,025
“Controllability means the ability of models to control the level of hallucination and strike a balance between faithfulness and diversity [41, 159].”
BackgroundMethodInfluentialCLIPScore: A Reference-free Evaluation Metric for Image Captioning
Hessel, J., Holtzman, A., Forbes, M., Le Bras, R., & Choi, Y. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.595 · Cited by 1,232
“Prior work has demonstrated that, for many literal description tasks, humans often prefer correctness in captions over specificity (Rohrbach et al., 2018, 2017).”
BackgroundDetecting hallucinations in large language models using semantic entropy
Farquhar, S., Kossen, J., Kuhn, L., & Gal, Y. (2024) · Nature · doi:10.1038/s41586-024-07421-0 · Cited by 934
A survey on multimodal large language models
Yin, S., Fu, C., Zhao, S., Li, K., Sun, X., Xu, T., & Chen, E. (2024) · National Science Review · doi:10.1093/nsr/nwae403 · Cited by 893
“In this paper, we perform a survey of the existing MLLM literature and offer a broad view of its main directions, including three common techniques (M-IT, M-ICL, and M-CoT) and a general framework to build task-solving systems (LAVR).”
Evaluating Object Hallucination in Large Vision-Language Models
Li, Y., Du, Y., Zhou, K., Wang, J., Xin, Z., & Wen, J.-R. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.20 · Cited by 1,862
From Show to Tell: A Survey on Deep Learning-Based Image Captioning
Stefanini, M., Cornia, M., Baraldi, L., Cascianelli, S., Fiameni, G., & Cucchiara, R. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3148210 · Cited by 373
“Another approach to captioning evaluation consists in relying on captions semantic similarity or other specific aspects of caption quality, which are estimated via embeddingbased metrics [158], [159], [160].”
BackgroundWang, T., Huang, J., Zhang, H., & Sun, Q. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.01077 · Cited by 291
RUBi: Reducing Unimodal Biases in Visual Question AnsweringPreprint
Cadène, R., Dancette, C., Ben-younes, H., Cord, M., & Parikh, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.10169 · Cited by 262
“In the language and vision context, some works evaluate unimodal baselines [37, 38] or leverage language priors [39].”
BackgroundMitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding
Leng, S., Zhang, H., Chen, G., Li, X., Lu, S., Miao, C., & Bing, L. (2024) · 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52733.2024.01316 · Cited by 199
“In the realm of VLMs, “object hal-lucination” is also well-documented, referring to models producing plausible outputs that include objects that do not match or are missing from images [6, 38, 54].”
BackgroundDeep Learning Approaches on Image Captioning: A Review
Ghandi, T., Pourreza, H. R., & Mahyar, H. (2023) · ACM Computing Surveys · doi:10.1145/3617592 · Cited by 219
“Object hallucination [ 104 ] is a persistent problem for image captioning models, wherein the model detects objects that are not present in the input image.”
BackgroundVisualGPT: Data-efficient Adaptation of Pretrained Language Models for Image Captioning
Chen, J., Guo, H., Yi, K., Li, B., & Elhoseiny, M. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01750 · Cited by 222
“In the second study, we evaluate if using pretrained language models introduces excessive linguistic prior that could cause the known object hallucination problem [62].”
BackgroundEvaluating the state-of-the-art of End-to-End Natural Language Generation: The E2E NLG challenge
Dušek, O., Novikova, J., & Rieser, V. (2019) · Computer Speech & Language · doi:10.1016/j.csl.2019.06.009 · Cited by 209
Zhou, L., Kalantidis, Y., Chen, X., Corso, J. J., & Rohrbach, M. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.00674 · Cited by 191
“While researchers have started to discover and study these problems for image description [14, 9, 24, 20],1 they are even more pronounced”
BackgroundInfluentialConnecting Vision and Language with Localized Narratives
Pont-Tuset, J., Uijlings, J., Changpinyo, S., Soricut, R., & Ferrari, V. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58558-7_38 · Cited by 188
Causal Attention for Vision-Language Tasks
Yang, X., Zhang, H., Qi, G., & Cai, J. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00972 · Cited by 175
Human evaluation of automatically generated text: Current trends and best practice guidelines
van der Lee, C., Albert Gatt, A., van Miltenburg, E., & Krahmer, E. (2020) · Computer Speech & Language · doi:10.1016/j.csl.2020.101151 · Cited by 208
Human-Centered Artificial Intelligence for Designing Accessible Cultural Heritage
Pisoni, G., Díaz-Rodríguez, N., Gijlers, H., & Tonolli, L. (2021) · Applied Sciences · doi:10.3390/app11020870 · Cited by 183
“The latter is a well-known phenomenon where image captioning models caption an image with objects not present in the image [122].”
Handling Divergent Reference Texts when Evaluating Table-to-Text Generation
Dhingra, B., Faruqui, M., Parikh, A., Chang, M.-W., Das, D., & Cohen, W. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1483 · Cited by 150
“Hallucination (Rohrbach et al., 2018; Lee et al., 2018) refers to when an NLG system generates text which mentions extra information than what is present in the source from which it is generated.”
BackgroundMore Grounded Image Captioning by Distilling Image-Text Matching Model
Zhou, Y., Wang, M., Liu, D., Hu, Z., & Zhang, H. (2020) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr42600.2020.00483 · Cited by 147
Reasoning with Language Model Prompting: A Survey
Qiao, S., Ou, Y., Zhang, N., Chen, X., Yao, Y., Deng, S., Tan, C., Huang, F., & Chen, H. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-long.294 · Cited by 346
Huang, Q., Dong, X., Zhang, P., Wang, B., He, C., Wang, J., Lin, D., Zhang, W., & Yu, N. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01274 · Cited by 270
“The Caption Hal-lucination Assessment with Image Relevance (CHAIR) [36] metric is a specifically crafted evaluation tool designed to assess object hallucination issues in image captioning task.”
Comprehensive Image Captioning via Scene Graph Decomposition
Zhong, Y., Wang, L., Chen, J., Yu, D., & Li, Y. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58568-6_13 · Cited by 140
Comprehending and Ordering Semantics for Image Captioning
Li, Y., Pan, Y., Yao, T., & Mei, T. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01746 · Cited by 139
“In addition, we use CHAIR metric [27] to assess the rate of object hallucination on the robust split.”
BackgroundMethodInfluentialDeconfounded Image Captioning: A Causal Retrospect
Yang, X., Zhang, H., & Cai, J. (2021) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2021.3121705 · Cited by 136
“Thus these captioners still easily learn the dataset bias [2].”
BackgroundMethodInfluentialInvariant Grounding for Video Question Answering
Li, Y., Wang, X., Xiao, J., Ji, W., & Chua, T. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00294 · Cited by 132
“Multi-modal datasets tend to display inherent bias in some forms [1, 19, 25, 28].”
BackgroundKnowledge-Augmented Language Model Prompting for Zero-Shot Knowledge Graph Question Answering
Baek, J., Aji, A. F., & Saffari, A. (2023) · NLRSE · doi:10.18653/v1/2023.nlrse-1.7 · Cited by 274
“However, since the knowledge in LLMs might be incomplete, incorrect, and outdated, they often generate factually wrong answers, known as hallucination (Rohrbach et al., 2018) (See Figure 1a).”
Logical Natural Language Generation from Open-Domain Tables
Chen, W., Chen, J., Su, Y., Chen, Z., & Wang, W. Y. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.708 · Cited by 142
Yu, T., Yao, Y., Zhang, H., He, T., Han, Y., Cui, G., Hu, J., Liu, Z., Zheng, H.-T., & Sun, M. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01310 · Cited by 267
“For trustworthiness, we perform evaluation on three benchmarks: (1) Object HalBench [34] is a widely adopted benchmark for assessing object hallucination in detailed image descriptions.”
MethodPerformance of large language models on a neurology board-style examination
Schubert, M. C., Wick, W., & Venkataramani, V. (2023) · JAMA Network Open · doi:10.1001/jamanetworkopen.2023.46721 · Cited by 115
Trustworthy journalism through AI
Opdahl, A. L., Tessem, B., Dang‐Nguyen, D., Motta, E., Setty, V., Throndsen, E., Tverberg, A., & Trattner, C. (2023) · Data & Knowledge Engineering · doi:10.1016/j.datak.2023.102182 · Cited by 165
Screen2Words: Automatic Mobile UI Summarization with Multimodal Learning
Wang, B. K., Li, G., Zhou, X., Chen, Z., Grossman, T., & Li, Y. (2021) · ACM Symposium on User Interface Software and Technology (UIST) · doi:10.1145/3472749.3474765 · Cited by 104
“objects that are not actually in a scene [38].”
BackgroundLearning to Collocate Neural Modules for Image Captioning
Yang, X., Zhang, H., & Cai, J. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00435 · Cited by 100
“Indeed, image captioning is not the only model that can easily exploit the dataset bias to captioning even without looking at the image [44], almost all existing models for visual reasoning tasks such as visual Q&A [23, 46, 48] have been spotted mode collapse to certain dataset idiosyncrasies and failed to reproduce t…”
BackgroundIshmam, M. F., Shovon, M. S. H., Mridha, M. F., & Dey, N. (2024) · Information Fusion · doi:10.1016/j.inffus.2024.102270 · Cited by 107
Adversarial Inference for Multi-Sentence Video Description
Park, J. S., Rohrbach, M., Darrell, T., & Rohrbach, A. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.00676 · Cited by 94
“methods have been shown to lead to unwanted artifacts, such as ungrammatical sentence endings [15], increased object hallucination rates [47] and lack of diverse content [36].”
MethodSelf-citationNeural data-to-text generation: A comparison between pipeline and end-to-end architectures
Ferreira, T. C., van der Lee, C., van Miltenburg, E., & Krahmer, E. (2019) · Tilburg University Research Portal · Cited by 107
QuestEval: Summarization Asks for Fact-based Evaluation
Scialom, T., Dray, P.-A., Lamprier, S., Piwowarski, B., Staiano, J., Wang, A., & Gallinari, P. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.529 · Cited by 175
“…while recent state-of-the-art generative models (Lewis et al., 2019; Zhang et al., 2019a) produce fluent summaries, they frequently contain false or unsupported information (Kry´sci´nski et al., 2019), a phenomenon also known as neural hallucination (Rohrbach et al., 2018; Zhao et al., 2020).”
BackgroundSemantic Noise Matters for Neural Natural Language Generation
Dušek, O., Howcroft, D. M., & Rieser, V. (2019) · Proceedings of the 12th International Conference on Natural Language Generation · doi:10.18653/v1/w19-8652 · Cited by 85
“…work on related tasks, which mostly reports on hallucinations (i.e. adding information not grounded in the input), as observed for image captioning (Rohrbach et al., 2018), sports report generation (Wiseman et al., 2017), machine translation (Koehn and Knowles, 2017; Lee et al., 2019), and…”
BackgroundResultExtractive Opinion Summarization in Quantized Transformer Spaces
Angelidis, S., Amplayo, R. K., Suhara, Y., Wang, X., & Lapata, M. (2021) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00366 · Cited by 85
“C L ] 8 D ec 2020 (Holtzman et al., 2020), hallucinations (Rohrbach et al., 2018), and the undesirable use of first-person narrative, a direct consequence of review-like generation.”
BackgroundNeural Path Hunter: Reducing Hallucination in Dialogue Systems via Path Grounding
Dziri, N., Madotto, A., Zaïane, O., & Bose, A. J. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.168 · Cited by 81
“, 2019), image captioning (Rohrbach et al., 2018), exposure bias (Wang and Sennrich, 2020), and question answering (Feng et al.”
BackgroundCalibrated Language Models Must Hallucinate
Kalai, A. T., & Vempala, S. S. (2024) · Proceedings of the 56th Annual ACM Symposium on Theory of Computing · doi:10.1145/3618260.3649777 · Cited by 91
Controlled Hallucinations: Learning to Generate Faithfully from Noisy Data
Filippova, K. (2020) · Findings of the Association for Computational Linguistics: EMNLP 2020 · doi:10.18653/v1/2020.findings-emnlp.76 · Cited by 79
On hallucination and predictive uncertainty in conditional language generation
Xiao, Y., & Wang, W. Y. (2021) · Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume · doi:10.18653/v1/2021.eacl-main.236 · Cited by 79
Are You Looking? Grounding to Multiple Modalities in Vision-and-Language Navigation
Hu, R., Fried, D., Rohrbach, A., Klein, D., Darrell, T., & Saenko, K. (2019) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/p19-1655 · Cited by 86
“Hendricks et al. (2018) and Rohrbach et al. (2018) find that neural captioning models often ground object mentions into incorrect objects due to correlations in the training data, and can hallucinate non-existing objects.”
BackgroundSelf-citationLee, Y., Shin, T. H., Tessier, L., Javidan, A. P., Jung, J. J., Hong, D., Strong, A. T., McKechnie, T., Malone, S., Jin, D., et al. (2024) · Surgery for Obesity and Related Diseases · doi:10.1016/j.soard.2024.03.011 · Cited by 79
Multi-Modal Hallucination Control by Visual Information Grounding
Favero, A., Zancato, L., Trager, M., Choudhary, S., Perera, P., Achille, A., Swaminathan, A. D., & Soatto, S. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01356 · Cited by 132
“We show that applying M3ID or DPO reduces the percentage of hallucinated objects in captioning tasks [20] by 25% and by 28%, respectively and improves accuracy on the POPE [9] VQA hallucination benchmark by 21% and 24% over the base model.”
Show, Deconfound and Tell: Image Captioning with Causal Inference
Liu, B., Wang, D., Yang, X., Zhou, Y. Z., Yao, R., Shao, Z., & Zhao, J. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01751 · Cited by 75
“This is due to the fact that the widelyadopted SCST optimization may cause biases in order to improve the CIDEr score [35].”
BackgroundMethodLi, Z., Wu, X., Du, H., Liu, F., Nghiem, H. T., & Shi, G. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00147 · Cited by 91
“…[129], GenAI-Bench [107] Hallucination Evaluates whether models are likely to hallucinate on certain visual and textual inputs HallusionBench [64], POPE [119], CHAIR [184], M-HalDetect [65], Hallu-Pi [45], Halle-Switch [240], BEAF [231], AutoHallusion [218], GAIVE [130], Hal-Eval [90], AMBER [203]”
Park, J., Oh, C., & Kim, H. Y. (2024) · Technology in Society · doi:10.1016/j.techsoc.2024.102705 · Cited by 86
TRUE: Re-evaluating Factual Consistency Evaluation
Honovich, O., Aharoni, R., Herzig, J., Taitelbaum, H., Kukliansy, D., Cohen, V., Scialom, T., Szpektor, I., Hassidim, A., & Matias, Y. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.287 · Cited by 113
Probing Image-Language Transformers for Verb Understanding
Hendricks, L. A., & Nematzadeh, A. (2021) · Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 · doi:10.18653/v1/2021.findings-acl.318 · Cited by 63
Self-citationA Token-level Reference-free Hallucination Detection Benchmark for Free-form Text Generation
Liu, T., Zhang, Y., Brockett, C., Mao, Y., Sui, Z., Chen, W., & Dolan, B. (2022) · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2022.acl-long.464 · Cited by 63
Iterative Edit-Based Unsupervised Sentence Simplification
Kumar, D., Mou, L., Golab, L., & Vechtomova, O. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.707 · Cited by 64
Shifting the Baseline: Single Modality Performance on Visual Navigation & QA
Thomason, J., Gordon, D., & Bisk, Y. (2019) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/n19-1197 · Cited by 63
“, 2015) as well as exposed misalignment between caption relevance and text-based metrics (Rohrbach et al., 2018).”
BackgroundTRUE: Re-evaluating Factual Consistency Evaluation
Honovich, O., Aharoni, R., Herzig, J., Taitelbaum, H., Kukliansy, D., Cohen, V., Scialom, T., Szpektor, I., Hassidim, A., & Matias, Y. (2022) · Proceedings of the Second DialDoc Workshop on Document-grounded Dialogue and Conversational Question Answering · doi:10.18653/v1/2022.dialdoc-1.19 · Cited by 56
“…text generation models for real-world applications is that they often generate factually inconsistent text with respect to the input they are conditioned on, or even completely “hallucinate” (Lee et al., 2018; Rohrbach et al., 2018; Maynez et al., 2020; Zhao et al., 2020) as exemplified in Table 1.”
BackgroundAccessible Cultural Heritage through Explainable Artificial Intelligence
Díaz-Rodríguez, N., & Pisoni, G. (2020) · ACM Conference on User Modeling, Adaptation and Personalization (UMAP) · doi:10.1145/3386392.3399276 · Cited by 55
“The latter is a well-known phenomenon where image captioning models caption an image with objects not present in the image [60].”
BackgroundKnowing What to Learn: A Metric-Oriented Focal Mechanism for Image Captioning
Ji, J., Ma, Y., Sun, X., Zhou, Y., Wu, Y., & Ji, R. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3183434 · Cited by 50
Joint Commonsense and Relation Reasoning for Image and Video Captioning
Hou, J., Wu, X., Zhang, X., Qi, Y., Jia, Y., & Luo, J. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i07.6731 · Cited by 50
“…these methods that directly extract explicit semantic concepts from external knowledge, our method uses external knowledge to reason relationships between semantic concepts via joint commonsense and relation reasoning, without facing the “hallucinating” problem as stated by (Rohrbach et al. 2018).”
MethodHalluciDoctor: Mitigating Hallucinatory Toxicity in Visual Instruction Data
Yu, Q., Li, J., Wei, L., Pang, L., Ye, W., Qin, B., Tang, S., Tian, Q., & Zhuang, Y. (2024) · 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52733.2024.01230 · Cited by 59
“3) attribute hallucination refers to inaccurate object properties in the description, such as size, color, and states.”
BackgroundKnowledge-Augmented Language Model Prompting for Zero-Shot Knowledge Graph Question Answering
Baek, J., Aji, A., & Saffari, A. (2023) · Proceedings of the First Workshop on Matching From Unstructured and Structured Data (MATCHING 2023) · doi:10.18653/v1/2023.matching-1.7 · Cited by 51
“However, since the knowledge in LLMs might be incomplete, incorrect, and outdated, they often generate factually wrong answers, known as hallucination (Rohrbach et al., 2018) (See Figure 1a).”
BackgroundMitigating Gender Bias in Captioning Systems
Tang, R., Du, M., Li, Y., Liu, Z., Zou, N., & Hu, X. (2021) · Proceedings of the Web Conference 2021 · doi:10.1145/3442381.3449950 · Cited by 50
“From the model architecture perspective, we consider models both with and without attention module where "attention" refers to models learn to pay attention to different image regions for caption word generation [30].”
BackgroundMethodInfluentialVisual Cluster Grounding for Image Captioning
Jiang, W., Zhu, M., Fang, Y., Shi, G., Zhao, X., & Liu, Y. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3177318 · Cited by 50
Efficient GPT-4V level multimodal large language model for deployment on edge devices
Yao, Y., Yu, T., Zhang, A., Wang, C., Cui, J., Zhu, H., Cai, T., Chen, C., Li, H., Zhao, W., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-61040-5 · Cited by 60
Cross-modal Coherence Modeling for Caption Generation
Alikhani, M., Sharma, P. K., Li, S., Soricut, R., & Stone, M. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.583 · Cited by 46
“Unfortunately, such dedicated annotation efforts cannot yield enough data for training robust generation models; the resulting generated captions are plagued by content hallucinations (Rohrbach et al., 2018; Sharma et al., 2018) that effectively preclude them for being used in real-world applications.”
BackgroundConsensus Graph Representation Learning for Better Grounded Image Captioning
Zhang, W., Shi, H., Tang, S., Xiao, J., Yu, Q., & Zhuang, Y. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i4.16452 · Cited by 48
“Previous studies (Rohrbach et al. 2018) believed that this caption hallucination problem is caused by biased or inappropriate visual-textual correlations learned from the datasets, i.”
BackgroundMethodPositive-Augmented Contrastive Learning for Image and Video Captioning Evaluation
Sarto, S., Barraco, M., Cornia, M., Baraldi, L., & Cucchiara, R. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00668 · Cited by 54
ShapeLLM: Universal 3D Object Understanding for Embodied Interaction
Qi, Z., Dong, R., Zhang, S., Geng, H., Han, C., Ge, Z., Yi, L., & Ma, K. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72775-7_13 · Cited by 102
PAIR: Planning and Iterative Refinement in Pre-trained Transformers for Long Text Generation
Hua, X., & Wang, L. (2020) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.57 · Cited by 48
Efficient Counterfactual Debiasing for Visual Question Answering
Kolling, C., Móre, M. D., Gavenski, N. S., Pooch, E., Parraga, O., & Barros, R. C. (2022) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv51458.2022.00263 · Cited by 42
“In multi-modal tasks, i.e. , tasks that combine language and visual information, several studies evaluate unimodal baselines [27] or rely on external knowledge to address biases [23].”
BackgroundExposing and Mitigating Spurious Correlations for Cross-Modal Retrieval
Kim, J. M., Koepke, A. S., Schmid, C., & Akata, Z. (2023) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw59228.2023.00257 · Cited by 41
“…language bias in VQA [8, 55], hubness in cross-modal retrieval [6], multimodal spurious correlations in VQA [1], the spuriousness in captioning [30], object hal-lucination in captioning [60], large-scale vision-language pretraining [74], or cross-modal retrieval in a unique e-commerce setting [47].”
BackgroundPaying More Attention to Image: A Training-Free Method for Alleviating Hallucination in LVLMs
Liu, S., Zheng, K., & Chen, W. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73010-8_8 · Cited by 104
Self-Supervised and Controlled Multi-Document Opinion Summarization
Elsahar, H., Coavoux, M., Rozen, J., & Gallé, M. (2021) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2021.eacl-main.141 · Cited by 42
Let there be a clock on the beach: Reducing Object Hallucination in Image Captioning
Biten, A. F., Gomez, L., & Karatzas, D. (2022) · 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) · doi:10.1109/wacv51458.2022.00253 · Cited by 43
Auto-Parsing Network for Image Captioning and Visual Question Answering
Yang, X., Gao, C., Zhang, H., & Cai, J. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00220 · Cited by 38
“All these scores are listed in Table 1 and we can find that our APN has the highest recalls of each kind of word and the lowest CHAIR scores.”
MethodHallucination of Multimodal Large Language Models: A Survey
Bai, Z., Wang, P., Xiao, T., He, T., Han, Z., Zhang, Z., & Shou, M. Z. (2024) · arXiv · doi:10.48550/arxiv.2404.18930 · Cited by 38
“CHAIR [90] .”
InfluentialExplain and improve: LRP-inference fine-tuning for image captioning models
Sun, J., Lapuschkin, S., Samek, W., & Binder, A. (2021) · Information Fusion · doi:10.1016/j.inffus.2021.07.008 · Cited by 52
“[27] points out that hallucinating less does not necessarily render higher sentence-level evaluation metrics [29], [31], [99], which is also in line with our observations in Table IV.”
BackgroundResultLearning to Generate Grounded Visual Captions Without Localization Supervision
Ma, C., Kalantidis, Y., AlRegib, G., Vajda, P., Rohrbach, M., & Kira, Z. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58523-5_21 · Cited by 35
“The resulting models, however, are known to have poor grounding performance [20], leading to undesirable behaviors (such as object hallucinations [27]), despite having high captioning accuracy.”
BackgroundProposed Framework for Human-like Language Processing of ChatGPT in Academic Writing
Mahyoob, M., Algaraady, J., & Alblwi, A. (2023) · International Journal of Emerging Technologies in Learning (iJET) · doi:10.3991/ijet.v18i14.41725 · Cited by 48
Transparent Human Evaluation for Image Captioning
Kasai, J., Sakaguchi, K., Dunagan, L., Morrison, J., Bras, R. J. L., Choi, Y., & Smith, N. A. (2022) · Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) · doi:10.18653/v1/2022.naacl-main.254 · Cited by 40
EMScore: Evaluating Video Captioning via Coarse-Grained and Fine-Grained Embedding Matching
Shi, Y., Yang, X., Xu, H., Yuan, C., Li, B., Hu, W., & Zha, Z.-J. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.01740 · Cited by 38
“(3) As pointed by [24], these reference-based metrics may under-penalize the captions with “hallucinating” descriptions since these metrics only measure similarity to references, and the visual relevance cannot be fully captured.”
BackgroundA Review of Transformer-Based Approaches for Image Captioning
Ondeng, O., Absaloms, H. O., & Akuon, P. O. (2023) · Applied Sciences · doi:10.3390/app131911103 · Cited by 39
Plausible May Not Be Faithful: Probing Object Hallucination in Vision-Language Pre-training
Dai, W., Liu, Z., Ji, Z., Su, D., & Fung, P. (2023) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2023.eacl-main.156 · Cited by 59
CLAIR: Evaluating Image Captions with Large Language Models
Chan, D. M., Petryk, S., Gonzalez, J. E., Darrell, T., & Canny, J. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.841 · Cited by 56
Self-citationVCoder: Versatile Vision Encoders for Multimodal Large Language Models
Jain, J., Yang, J., & Shi, H. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.02644 · Cited by 53
“Despite the availability of various metrics [37, 45, 59] to measure object hallucination in vision-language models, no existing metric considers the explicit object counts while calculating their hallucination scores.”
BackgroundMethodInfluentialWith a Little Help from your own Past: Prototypical Memory Networks for Image Captioning
Barraco, M., Sarto, S., Cornia, M., Baraldi, L., & Cucchiara, R. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00282 · Cited by 30
“When evaluating our results on robust COCO, we also employ the CHAIR metric [43] that measures which fraction of objects mentioned in the generated sentences is hallucinated (CHi) and the portion of sentences that includes a hallucinated object (CHs).”
MethodWang, L., He, J., Li, S., Liu, N., & Lim, E. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-53302-0_3 · Cited by 38
Towards Unique and Informative Captioning of Images
Wang, Z., Feng, B. T., Narasimhan, K., & Russakovsky, O. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58571-6_37 · Cited by 29
Hybrid attention network for image captioning
Jiang, W., Li, Q., Zhan, K., Fang, Y., & Shen, F. (2022) · Displays · doi:10.1016/j.displa.2022.102238 · Cited by 29
Benchmark Evaluations, Applications, and Challenges of Large Vision Language Models: A SurveyPreprint
Li, Z., Wu, X., Du, H., Nghiem, H. T., & Shi, G. (2025) · Qeios · doi:10.32388/gxr68q · Cited by 33
Relational Graph Learning for Grounded Video Description Generation
Zhang, W., Wang, X. E., Tang, S., Shi, H., Shi, H., Xiao, J., Zhuang, Y., & Wang, W. Y. (2020) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3394171.3413746 · Cited by 29
“According to [23], we compute CHAIR and CHAIR , which indicate the fraction of object instances that are hallucinated per-instance and per-sentence, respectively.”
BackgroundMethodCAST: Cross-Modal Retrieval and Visual Conditioning for image captioning
Cao, S., An, G., Cen, Y., Yang, Z., & Lin, W. (2024) · Pattern Recognition · doi:10.1016/j.patcog.2024.110555 · Cited by 28
Deconfounded Image Captioning: A Causal RetrospectPreprint
Yang, X., Zhang, H., & Cai, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.03923 · Cited by 43
“Thus these captioners still easily learn the dataset bias [2].”
BackgroundMethodInfluentialAlAfnan, M. A. (2024) · Journal of Artificial Intelligence and Technology · doi:10.37965/jait.2024.0549 · Cited by 45
Large multimodal models evaluation: a survey
Zhang, Z., Wang, J., Wen, F., Guo, Y., Zhao, X., Fang, X., Ding, S., Jia, Z., Xiao, J., Shen, Y., et al. (2025) · Science China Information Sciences · doi:10.1007/s11432-025-4676-4 · Cited by 39
“Early studies such as Rohrbach et al. [214] analyzed object hallucination in image captioning, laying the foundation for systematic evaluation.”
QuestEval: Summarization Asks for Fact-based EvaluationPreprint
Scialom, T., Dray, P.-A., Gallinari, P., Lamprier, S., Piwowarski, B., Staiano, J., & Wang, A. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.12693 · Cited by 22
“…while recent state-of-the-art generative models (Lewis et al., 2019; Zhang et al., 2019a) produce fluent summaries, they frequently contain false or unsupported information (Kry´sci´nski et al., 2019), a phenomenon also known as neural hallucination (Rohrbach et al., 2018; Zhao et al., 2020).”
BackgroundEvidence-based Factual Error Correction
Thorne, J., & Vlachos, A. (2021) · Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) · doi:10.18653/v1/2021.acl-long.256 · Cited by 39
“accurate, or supported by the retrieved evidence, commonly referred to as a hallucination (Rohrbach et al., 2018; Zhou et al., 2020).”
BackgroundFusion Models for Improved Image Captioning
Kalimuthu, M., Mogadala, A., Mosbach, M., & Klakow, D. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-68780-9_32 · Cited by 21
LUNA: A Model-Based Universal Analysis Framework for Large Language Models
Song, D., Xie, X., Song, J., Zhu, D., Huang, Y., Juefei-Xu, F., & Ma, L. (2024) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2024.3411928 · Cited by 33
“3) Hallucination: With the advancement of language models, some research works have posed these models can occasionally produce unfaithful and nonfactual content considering the given inputs [113], [114], [115] and such undesirable generation behavior is so-called hallucination [15].”
BackgroundChen, Y., Sikka, K., Cogswell, M., Ji, H., & Divakaran, A. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01350 · Cited by 56
“We evaluate the honesty (a.k.a, hallucination control) of DRESS using the image captioning task following [12, 56, 89].”
MethodAn, W., Tian, F., Leng, S., Nie, J., Lin, H., Wang, Q., Chen, P., Zhang, X., & Lu, S. (2025) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52734.2025.02784 · Cited by 77
“The improved performance [52].”
InfluentialVolcano: Mitigating Multimodal Hallucination through Self-Feedback Guided Revision
Lee, S., Park, S. H., Jo, Y., & Seo, M. (2024) · Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) · doi:10.18653/v1/2024.naacl-long.23 · Cited by 20
Metadata
- Authors
- Rohrbach, A., Hendricks, L. A., Burns, K., Darrell, T., & Saenko, K.
- Affiliations
- Max Planck Society (DE)
- University of California, Berkeley (US)
- Boston University (US)
- Venue
- Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing
- Published
- 2018
- Pages
- 4035-4045
- OpenAlex
- W2892087592
- Cited by
- 420
- References
- 19
- Citations by year
- 2026: 103 · 2025: 80 · 2024: 70 · 2023: 42 · 2022: 32 · 2021: 45 · 2020: 32 · 2019: 15 · 2018: 1
- FWCI
- 7.01
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference on Empirical Methods in Natural Language Processing (EMNLP) (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Keywords
- object hallucination, image captioning, hallucination evaluation, language priors
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
