Foundation Models Defining a New Era in Vision: A Survey and Outlook
The DOI matches its Crossref record; 227 of the 227 references it cites were found in a registry or an index.
Abstract
Vision systems that see and reason about the compositional nature of visual scenes are fundamental to understanding our world. The complex relations between objects and their locations, ambiguities, and variations in the real-world environment can be better described in human language, naturally governed by grammatical rules and other modalities such as audio and depth. The models learned to bridge the gap between such modalities and large-scale training data facilitate contextual reasoning, generalization, and prompt capabilities at test time. These models are referred to as foundation models. The output of such models can be modified through human-provided prompts without retraining, e.g., segmenting a particular object by providing a bounding box, having interactive dialogues by asking questions about an image or video scene or manipulating the robot's behavior through language instructions. In this survey, we provide a comprehensive review of such emerging foundation models, including typical architecture designs to combine different modalities (vision, text, audio, etc.), training objectives (contrastive, generative), pre-training datasets, fine-tuning mechanisms, and the common prompting patterns; textual, visual, and heterogeneous. We discuss the open challenges and research directions for foundation models in computer vision, including difficulties in their evaluations and benchmarking, gaps in their real-world understanding, limitations of contextual understanding, biases, vulnerability to adversarial attacks, and interpretability issues. We review recent developments in this field, covering a wide range of applications of foundation models systematically and comprehensively.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
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?
In the citing paper
“Vision Foundation Models (VFMs)-large-scale models pre-trained on massive collections of unlabeled images to learn robust and transferable visual representations [5,6]”
From Trustworthy Celestial Eye: Calibrated and Robust Planetary Classification via Self-Supervised Vision Transformers (2026) · cites this paper as [5]
Citations
Cites
Items 201–227 of 227
(2023) · doi:10.52202/075280-2687 · Cited by 57
CLIP the Bias: How Useful is Balancing Data in Multimodal Learning?
Alabdulmohsin, I., Wang, X., Steiner, A., Goyal, P., D’Amour, A., & Zhai, X. (2024) · International Conference on Learning Representations · doi:10.48550/arxiv.2403.04547 · Cited by 55 · cited 1×
“Moreover, recent studies have investigated issues of fairness and robustness in CLIP-like foundational models [80], [81], [82].”
Foundation TransformersPreprint
Wang, H., Ma, S., Huang, S., Li, D., Wang, W., Peng, Z., Wu, Y., Bajaj, P., Singhal, S., Benhaim, A., et al. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2210.06423 · Cited by 54 · cited 1×
“The KOSMOS-1 model consists of a Magneto-based Large Language Model (LLM) [109] as a general interface and xPos [139] encoders to encode different modalities.”
Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation
Mündler, N., He, J., Jenko, S., & Vechev, M. (2023) · arXiv · doi:10.48550/arxiv.2305.15852 · Cited by 47 · cited 1×
“Strategies tomitigatehallucinationincludeprovidingexplicitinstructions (orso-called system commands ), the chain of thought prompting [175], [176], [177], self-consistency (voting) [178], [179] and use of knowledge bases for retrieval augmented generation [180], [181], [182].”
Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation
Mündler, N., He, J., Jenko, S., & Vechev, M. (2023) · arXiv (Cornell University) · doi:10.3929/ethz-b-000719979 · Cited by 47
ViewRefer: Grasp the Multi-view Knowledge for 3D Visual Grounding
Guo, Z., Tang, Y., Zhang, R., Wang, D., Wang, Z., Zhao, B., & Li, X. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01410 · Cited by 44
On the Robustness of Segment AnythingPreprint
Huang, Y., Cao, Y., Li, T., Juefei-Xu, F., Lin, D., Tsang, I. W.-H., Liu, Y., & Guo, Q. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2305.16220 · Cited by 43 · cited 1×
“Visual prompt-based models, e.g., SAM, have also been attacked by corrupting their inputs and associated prompts [212], [213].”
Hallucination 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 · cited 1×
“Curious readers may refer to [191] for a more comprehensive treatment of the subject.”
Vision-Language Models for Vision Tasks: A SurveyPreprint
Zhang, J., Huang, J., Jin, S., et al. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2304.00685 · Cited by 37 · cited 1×
“The work of [28], [29] summarizes different vision-language pre-training network architectures, objectives, and downstream tasks and catego-rizes vision-language pre-training frameworks.”
RefSAM: Efficiently Adapting Segmenting Anything Model for Referring Video Object SegmentationPreprint
Li, Y., Zhang, J., Teng, X., Lan, L., & Liu, X. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2307.00997 · Cited by 36
Retrieval Augmented Chest X-Ray Report Generation using OpenAI GPT modelsPreprint
Ranjit, M., Ganapathy, G., Manuel, R. F., & Ganu, T. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2305.03660 · Cited by 33 · cited 1×
“Strategies tomitigatehallucinationincludeprovidingexplicitinstructions (orso-called system commands ), the chain of thought prompting [175], [176], [177], self-consistency (voting) [178], [179] and use of knowledge bases for retrieval augmented generation [180], [181], [182].”
Wang, J., Zhang, Y., & Sang, J. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2210.14562 · Cited by 30 · cited 1×
“Moreover, recent studies have investigated issues of fairness and robustness in CLIP-like foundational models [80], [81], [82].”
BAPLe: Backdoor Attacks on Medical Foundational Models Using Prompt Learning
Hanif, A., Shamshad, F., Awais, M., Naseer, M., Khan, F. S., Nandakumar, K., Khan, S. H., & Anwer, R. M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72390-2_42 · Cited by 28
Self-citationRetrieving-to-Answer: Zero-Shot Video Question Answering with Frozen Large Language Models
Pan, J., Lin, Z., Ge, Y., Zhu, X., Zhang, R., Wang, Y., Qiao, Y., & Li, H. (2023) · 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW) · doi:10.1109/iccvw60793.2023.00035 · Cited by 27 · cited 1×
“Strategies tomitigatehallucinationincludeprovidingexplicitinstructions (orso-called system commands ), the chain of thought prompting [175], [176], [177], self-consistency (voting) [178], [179] and use of knowledge bases for retrieval augmented generation [180], [181], [182].”
LLaVAR: Enhanced Visual Instruction Tuning for Text-Rich Image Understanding
Zhang, Y., Zhang, R., Gu, J., Zhou, Y., Lipka, N., Yang, D., & Sun, T. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2306.17107 · Cited by 27 · cited 1×
“Similarly, Zhang et al. [163] also present a visual-tuning dataset.”
Mitigating Hallucination in Large Multi-Modal Models via Robust Instruction Tuning
Liu, F., Lin, K., Li, L., Wang, J., Yacoob, Y., & Wang, L. (2023) · arXiv (Cornell University) · doi:10.13140/rg.2.2.32487.42402 · Cited by 23
Embodied Task Planning with Large Language Models
Wu, Z., Wang, Z., Xu, X., Lu, J., & Yan, H. (2023) · arXiv · doi:10.48550/arxiv.2307.01848 · Cited by 23
Zhang, X., Zeng, Y., Zhang, J., & Li, H. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.40 · Cited by 21
MixACM: Mixup-Based Robustness Transfer via Distillation of Activated Channel MapsPreprint
Awais, M., Zhou, F., Xie, C., Li, J., Bae, S., & Li, Z. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.05073 · Cited by 20 · cited 1×
“These attacks involve carefully crafted inputs that can cause the model to generate incorrect or harmful outputs [201], [202], [203].”
Self-citationSkreta, M., Yoshikawa, N., Arellano-Rubach, S., Ji, Z., Kristensen, L. B., Darvish, K., Aspuru-Guzik, A., Shkurti, F., & Garg, A. (2023) · arXiv · doi:10.48550/arxiv.2303.14100 · Cited by 18
eP-ALM: Efficient Perceptual Augmentation of Language Models
Shukor, M., Dancette, C., & Cord, M. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.02016 · Cited by 16
COSA: Concatenated Sample Pretrained Vision-Language Foundation ModelPreprint
Chen, S., He, X., Li, H., Jin, X., Feng, J., & Liu, J. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2306.09085 · Cited by 15 · cited 1×
“The second category includes approaches like MACAW-LLM [19], an instruction-tunedmulti-modalLLMthatintegratesfourdifferentmodalities, COSA [155], which aligns LLMs with videos, and Valley [34], a multi-modal framework capable of integrating video, image, and language perceptions.”
Can GPT-4 Perform Neural Architecture Search?
Zheng, M., Su, X., You, S., Wang, F., Qian, C., Xu, C., & Albanie, S. (2023) · arXiv · doi:10.48550/arxiv.2304.10970 · Cited by 15
RETA-LLM: A Retrieval-Augmented Large Language Model Toolkit
Liu, J., Jin, J., Wang, Z., Cheng, J., Dou, Z., & Wen, J.-R. (2023) · arXiv · doi:10.48550/arxiv.2306.05212 · Cited by 13 · cited 1×
“Strategies tomitigatehallucinationincludeprovidingexplicitinstructions (orso-called system commands ), the chain of thought prompting [175], [176], [177], self-consistency (voting) [178], [179] and use of knowledge bases for retrieval augmented generation [180], [181], [182].”
TaCA: Upgrading Your Visual Foundation Model with Task-agnostic Compatible AdapterPreprint
Zhang, B., Ge, Y., Xu, X., Shan, Y., & Shou, M. Z. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2306.12642 · Cited by 8 · cited 1×
“Furthermore, in model enhancement, Zhang et al. [129] present an efficient framework for seamlessly upgrading legacy foundation models to new tasks.”
Stone Needle: A General Multimodal Large-scale Model Framework towards HealthcarePreprint
Liu, W., & Zuo, Y. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2306.16034 · Cited by 6 · cited 1×
“4) and contrastive methods from different perspectives, such as remote sensing [77], visual document understanding [78] and health-care [79].”
(2025) · doi:10.24894/sprengglossarium_b03425 · Cited by 2
Research works printed but not linked: 10
- Radford, 2019, Language models are unsupervised multitask learners, OpenAI Blog, 1
- Schuhmann, LAION-5B: An open large-scale dataset for training next generation image-text models, 25278
- 10.5555/3524938.3525087 · doi:10.5555/3524938.3525087
- Liu, 2019, RoBERTa: A robustly optimized bert pretraining approach, ArXiv
- Zhang, GLIPv2: Unifying localization and vision-language understanding, 36067
- Kwon, Masked vision and language modeling for multi-modal representation learning, 4
- Ordonez, Im2text: Describing images using 1 million captioned photographs, 3
- 2023, GPT-4 technical report, PREPRINT
- 10.18653/v1/2023.emnlp-main.20 · DOI · Paper page
- Hu, LoRA: Low-rank adaptation of large language models, 3
Cited by
Segment anything model for medical image segmentation: Current applications and future directions
Zhang, Y., Shen, Z., & Jiao, R. (2024) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2024.108238 · Cited by 332
Foundation Models for Time Series Analysis: A Tutorial and Survey
Liang, Y., Wen, H., Nie, Y., Jiang, Y., Jin, M., Song, D., Pan, S., & Wen, Q. (2024) · Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining · doi:10.1145/3637528.3671451 · Cited by 225
GLaMM: Pixel Grounding Large Multimodal Model
Rasheed, H., Maaz, M., Shaji, S., Shaker, A., Khan, S., Cholakkal, H., Anwer, R. M., Xing, E., Yang, M.-H., & Khan, F. S. (2024) · 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52733.2024.01236 · Cited by 189
Self-citationHadi, M. U., tashi, Q. A., Qureshi, R., Shah, A., muneer, A., Irfan, M., Zafar, A., Shaikh, M. B., Akhtar, N., Wu, J., et al. (2023) · doi:10.36227/techrxiv.23589741.v4 · Cited by 103
Self-citationVSCode: General Visual Salient and Camouflaged Object Detection with 2D Prompt Learning
Luo, Z., Liu, N., Zhao, W., Yang, X., Zhang, D., Fan, D.-P., Khan, F. S., & Han, J. (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/cvpr52733.2024.01625 · Cited by 237
Self-citation
Metadata
- Authors
- Awais, M., Naseer, M., Khan, S., Anwer, R. M., Cholakkal, H., Shah, M., Yang, M.-H., & Khan, F. S.
- Affiliations
- MBZ University of AI, Abu Dhabi, UAE
- Australian National University (AU)
- Mohamed bin Zayed University of Artificial Intelligence (AE)
- University of Central Florida (US)
- University of California, Merced (US)
- Linköping University (SE)
- ORCID
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2025-04
- Volume
- 47
- Issue
- 4
- Pages
- 2245-2264
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- 1939-3539
- OpenAlex
- W4406208390
- Cited by
- 304
- References
- 198
- Citations by year
- 2026: 184 · 2025: 118
- FWCI
- 105.44
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- 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 › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Keywords
- vision foundation models, multimodal models, vision-language models, prompting, pre-training dataset, fine-tuning, contrastive learning, generative models, benchmarking, adversarial attacks, interpretability
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
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.