Self-Supervised Visual Feature Learning With Deep Neural Networks: A Survey
Open accessThe DOI matches its Crossref record; 204 of the 206 references it cites were found in a registry or an index.
Abstract
Large-scale labeled data are generally required to train deep neural networks in order to obtain better performance in visual feature learning from images or videos for computer vision applications. To avoid extensive cost of collecting and annotating large-scale datasets, as a subset of unsupervised learning methods, self-supervised learning methods are proposed to learn general image and video features from large-scale unlabeled data without using any human-annotated labels. This paper provides an extensive review of deep learning-based self-supervised general visual feature learning methods from images or videos. First, the motivation, general pipeline, and terminologies of this field are described. Then the common deep neural network architectures that used for self-supervised learning are summarized. Next, the schema and evaluation metrics of self-supervised learning methods are reviewed followed by the commonly used datasets for images, videos, audios, and 3D data, as well as the existing self-supervised visual feature learning methods. Finally, quantitative performance comparisons of the reviewed methods on benchmark datasets are summarized and discussed for both image and video feature learning. At last, this paper is concluded and lists a set of promising future directions for self-supervised visual feature learning.
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
“Jing and Tian [18] summarize the self-supervised feature learning methods on image and video data, and four categories are discussed: generation-based, context-based, free semantic label-based, and cross modal-based, where cross-modal-based methods construct learning task using RGB frame sequence an optical flow sequence, which are unique features in the video.”
From Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and Prospects (2024) · cites this paper as [18]
In the citing paper
“Another hot research sub-area is self-supervised learning, which learns discriminative visual representation by solving predefined pretext tasks [194].”
From Empowering things with intelligence: a survey of the progress, challenges, and opportunities in artificial intelligence of things (2021) · cites this paper as [194]
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
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 10×
“Figure is reproduced based on VGG [9].”
BackgroundMethodInfluentialImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 8×
“Trained with accurate human-annotated labels, the supervised learning methods obtained break-through results on different computer vision applications [1], [4], [8], [16].”
MethodInfluentialHochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 1×
“Compared to standard RNN [72], long short term memory (LSTM) uses memory cells to store, modify, and access internal states, to better model the long-term temporal relationships within video frames [73].”
BackgroundFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2577031 · Cited by 73,472 · cited 2×
“DUE to the powerful ability to learn different levels of general visual features, deep neural networks have been used as the basic structure to many computer vision applications such as object detection [1], [2], [3], semantic segmentation [4], [5], [6], image captioning [7], etc.”
BackgroundMethodFaster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
Ren, S., He, K., Girshick, R., & Sun, J. (2015) · arXiv · doi:10.48550/arxiv.1506.01497 · Cited by 70,812 · cited 2×
“DUE to the powerful ability to learn different levels of general visual features, deep neural networks have been used as the basic structure to many computer vision applications such as object detection [1], [2], [3], semantic segmentation [4], [5], [6], image captioning [7], etc.”
BackgroundMethodImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 5×
“• ImageNet: The ImageNet dataset [13] contains 1.”
BackgroundInfluentialGradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 1×
“• MNIST: The MNIST is a dataset of handwritten digits consisting of 70, 000 images while 60, 000 images belong to training set and the rest 10, 000 images are for testing [110].”
BackgroundDistinctive Image Features from Scale-Invariant Keypoints
Lowe, D. G. (2004) · International Journal of Computer Vision · doi:10.1023/b:visi.0000029664.99615.94 · Cited by 55,589
You Only Look Once: Unified, Real-Time Object Detection
Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.91 · Cited by 53,803 · cited 1×
“Recently, many datasets such as MSCOCO [99] and OpenImage [14] have been proposed for object detection and many ConvNet-based models [1], [2], [3], [100], [101], [102], [103], [104] have been proposed and obtained great performance.”
Background
Cited by
Items 601–700 of 1,005
In Defense of Lazy Visual Grounding for Open-Vocabulary Semantic Segmentation
Kang, D., & Cho, M. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72940-9_9 · Cited by 37
Said, Y., Boubaker, S., Altowaijri, S. M., Alsheikhy, A. A., & Atri, M. (2025) · Mathematics · doi:10.3390/math13060909 · Cited by 26
Self-supervised representation learning for geospatial objects: A survey
Chen, Y., Huang, W., Zhao, K., Jiang, Y., & Cong, G. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.103265 · Cited by 33
“Related Surveys and Our Distinction: Several recent surveys have discussed the application of SSL, mainly focusing on other domains, such as general SSL [11], SSL for computer vision [13], graphs [17, 18], time series [19], and recommender systems [22].”
Chiaroni, F., Rahal, M.-C., Hueber, N., & Dufaux, F. (2020) · IEEE Signal Processing Magazine · doi:10.1109/msp.2020.2977269 · Cited by 24
“SSL for autonomous driving vehicles perception is most often based on learning from data that is automatically labeled by an upstream method, similar to feature learning in [14].”
MethodSrinidhi, C. L., & Martel, A. L. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00069 · Cited by 32
“, pretext task) to pretrain the model on the unlabeled data [22].”
BackgroundTwo-Stage Self-supervised Cycle-Consistency Network for Reconstruction of Thin-Slice MR Images
Lu, Z., Li, Z., Wang, J., Shi, J., & Shen, D. G. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-87231-1_1 · Cited by 29
The Coming of Age of Interpretable and Explainable Machine Learning Models
Lisboa, P. J. G., Saralajew, S., Vellido, A., & Villmann, T. (2021) · European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN) · doi:10.14428/esann/2021.es2021-2 · Cited by 27
“Based on a contrastive loss, these models try to either embed the data into meaningful manifolds of low dimensionality, or to extract interpretable features, which can serve as input for other models [27, 28, 37].”
BackgroundSSL++: Improving Self-Supervised Learning by Mitigating the Proxy Task-Specificity Problem
Song, C., Xue, J., Chang, J., Zhang, J., Yang, J., & Tian, Q. C. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3135470 · Cited by 25
“In addition to designing various proxy tasks, the difficulty of proxy tasks is also an important factor [4].”
BackgroundMethodInfluentialSimilarity learning for wells based on logging data
Romanenkova, E., Rogulina, A., Shakirov, А., Stulov, N., Zaytsev, A., Ismailova, L., Kovalev, D., Katterbauer, K., & Alshehri, A. A. (2022) · Journal of Petroleum Science and Engineering · doi:10.1016/j.petrol.2022.110690 · Cited by 24
Radar Intra–Pulse Signal Modulation Classification with Contrastive Learning
Cai, J., Gan, F., Cao, X., Liu, W., & Li, P. (2022) · Remote Sensing · doi:10.3390/rs14225728 · Cited by 23
House Price Prediction Using Exploratory Data Analysis and Machine Learning with Feature Selection
Basysyar, F. M., & Dwilestari, G. (2022) · Acadlore Transactions on AI and Machine Learning · doi:10.56578/ataiml010103 · Cited by 23
Unsupervised Learning of Disentangled Representation via Auto-Encoding: A Survey
Eddahmani, I., Pham, C.-H., Napoléon, T., Badoc, I., Fouefack, J.-R., & Elbouz, M. (2023) · Sensors · doi:10.3390/s23042362 · Cited by 27
A survey on preprocessing and classification techniques for acoustic scene
Singh, V. K., Sharma, K., & Sur, S. N. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.120520 · Cited by 23
Pseudo Features-Guided Self-Training for Domain Adaptive Semantic Segmentation of Satellite Images
Zhang, F., Shi, Y., Xiong, Z., Huang, W., & Zhu, X. X. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3281503 · Cited by 25
“Since self-supervised learning has achieved remarkable progress [43], the importance of data augmentation has been widely acknowledged.”
BackgroundPedestrian Crossing Action Recognition and Trajectory Prediction with 3D Human Keypoints
Li, J., Shi, X., Chen, F., Stroud, J., Zhang, Z., Lan, T., Mao, J., Kang, J., Refaat, K. S., Yang, W., et al. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160273 · Cited by 37
“More specifically, we generalize the pretext tasks in the context of self-supervised learning that are designed to learn spatial patterns of images to capture temporal patterns of the keypoints sequences [12]–[19].”
BackgroundSelf-Supervised Interactive Embedding for One-Shot Organ Segmentation
Yang, Y., Wang, B., Zhang, D., Yuan, Y., Yan, Q., Zhao, S., You, Z., & Han, J. (2023) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2023.3265033 · Cited by 24
Qu, Y., Li, J., Huang, X., & Wen, D. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3319961 · Cited by 24
“The way to construct sample pairs is one of the key points of self-supervised learning [23].”
BackgroundSAR image generation method for oriented ship detection via generative adversarial networks
Ju, M., Niu, B., & Zhang, J. (2023) · Signal Image and Video Processing · doi:10.1007/s11760-023-02793-8 · Cited by 23
Zhang, Q., & Hou, J. (2023) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2023.3345353 · Cited by 27
Solving partial differential equations using large-data models: a literature review
Hafiz, A. M., Faiq, I., & Hassaballah, M. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10784-5 · Cited by 29
“CNNs pre-trained on large datasets like Ima-geNet (Deng et al. 2009) are used after fine-tuning for two notable reasons (Jing and Tian 2021).”
Enhancing skin lesion diagnosis with data augmentation techniques: a review of the state-of-the-art
Patil, A. Y., Mehto, A., & Nalband, S. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-20145-7 · Cited by 32
“Self-supervisedlearning,whichisrelatedtogenerativemodels,isthestudyofbuilding strongintermediaterepresentationsthatcanbeusedinlaterunsupervisedtasks[62].”
DED-SAM:Adapting Segment Anything Model 2 for Dual Encoder–Decoder Change Detection
Qiu, J., Liu, W., Zhang, X., Li, E., Zhang, L., & Li, X. (2024) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2024.3490754 · Cited by 23
An Evaluation of Self-supervised Pre-training for Skin-Lesion Analysis
Chaves, L., Bissoto, A. E., Junior, E. A. D. V., & Avila, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-25069-9_11 · Cited by 29
Tu, Y., Lei, H., Shen, H., & Yang, Y. (2021) · Briefings in Bioinformatics · doi:10.1093/bib/bbab605 · Cited by 23
Large-Scale Hyperspectral Image Clustering Using Contrastive LearningPreprint
Cai, Y., Zhang, Z., Liu, Y., Ghamisi, P., Li, K., Liu, X., & Cai, Z. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.07945 · Cited by 28
“Self-supervised learning [38], as a subset of unsupervised learning, has emerged as a promising paradigm for representation learning.”
Pre-trained artificial intelligence-aided analysis of nanoparticles using the segment anything model
Monteiro, G. A. A., Monteiro, B. A. A., dos Santos, J. A., & Wittemann, A. (2025) · Scientific Reports · doi:10.1038/s41598-025-86327-x · Cited by 25
RP-DETR: end-to-end rice pests detection using a transformer
Wang, J., Wang, T., Xu, Q., Gao, L., Gu, G., Jia, L., & Yao, C. (2025) · Plant Methods · doi:10.1186/s13007-025-01381-w · Cited by 22
“Deep learning features refer to the visual feature expression established for an image through a deep neural network model [19].”
A survey on self-supervised learning: Recent advances and open problems
Zhang, J., Yang, L., Mohammadabadi, S. M. S., & Yan, F. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.131409 · Cited by 32
A Low Rank Promoting Prior for Unsupervised Contrastive Learning
Wang, Y., Lin, J., Cai, Q., Pan, Y., Yao, T., Chao, H., & Mei, T. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3180995 · Cited by 24
Surrogate- and invariance-boosted contrastive learning for data-scarce applications in science
Loh, C., Christensen, T., Dangovski, R., Kim, S., & Soljacic, M. (2022) · Nature Communications · doi:10.1038/s41467-022-31915-y · Cited by 26
“Extending from TL, self-supervised learning (SSL) [27, 28] is a technique where the pre-training stage uses only unlabeled data.”
BackgroundAttention-Driven Appearance-Motion Fusion Network for Action Recognition
Liu, S., & Ma, X. (2022) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2022.3148588 · Cited by 22
“In the past decades, a lot of advanced works have been done on human action recognition in videos [2].”
BackgroundDeep semi-supervised learning with contrastive learning and partial label propagation for image data
Gan, Y., Zhu, H., Guo, W., Xu, G., & Zou, G. (2022) · Knowledge-Based Systems · doi:10.1016/j.knosys.2022.108602 · Cited by 23
Dual-Distribution Discrepancy for Anomaly Detection in Chest X-Rays
Cai, Y., Chen, H., Yang, X. Y., Zhou, Y., & Cheng, K.-T. T. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-16437-8_56 · Cited by 32
SWIN transformer based contrastive self-supervised learning for animal detection and classification
Agilandeeswari, L., & Sundaram, D. M. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-13629-x · Cited by 25
“Pretext tasks [19] are a variety of tasks that predict from the input without labels and aid in acquiring an effective representation for the pertinent downstream task.”
BackgroundCurrent Advances in Computational Lung Ultrasound Imaging: A Review
Yang, T., Karakuş, O., Anantrasirichai, N., & Achim, A. (2022) · IEEE Transactions on Ultrasonics Ferroelectrics and Frequency Control · doi:10.1109/tuffc.2022.3221682 · Cited by 25
“This is where self-supervised methods play a vital tole in fuelling the progress of DL without the need for expensive annotations and learn feature representations where data provide supervision [80].”
MethodTracking Passengers and Baggage Items Using Multiple Overhead Cameras at Security Checkpoints
Siddique, A., & Medeiros, H. P. (2022) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2022.3225252 · Cited by 25
“Previous works have used SSL techniques to improve visual feature learning [39]–[41], reducing dependency on human annotations for training backbone models.”
MethodSIR: Self-Supervised Image Rectification via Seeing the Same Scene From Multiple Different Lenses
Fan, J., Zhang, J., & Tao, D. (2023) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3231087 · Cited by 26
“ifferent distortion parameters, thereby improving the generalization. Self-supervised Learning. As a sub-area of unsupervised learning, self-supervised learning has been a hot research topic recently [16,26,17]. It aims to learn a useful feature representation for downstream tasks by solving the pretext tasks. The pre…”
BackgroundLuo, G., Zhang, H., Yuan, Q., Li, J., Wang, W., & Wang, F. (2023) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3259045 · Cited by 22
“From the perspective of representative learning [66], by going deeper, it is beneficial to extract compact and semantic-rich features that are beneficial to predict the complex and nonlinear correlations.”
BackgroundDeep Learning in Environmental Toxicology: Current Progress and Open Challenges
Tan, H., Jin, J., Fang, C., Zhang, Y., Chang, B., Zhang, X., Yu, H., & Shi, W. F. (2023) · ACS ES&T Water · doi:10.1021/acsestwater.3c00152 · Cited by 26
An overview of ensemble and feature learning in few-shot image classification using siamese networks
Valero-Mas, J. J., Gallego, A., & Rico-Juan, J. R. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-15607-3 · Cited by 28
“For these experiments, the Euclidean distance is used as dissimilarity metric and the influence of parameter k is assessed in the range k ∈ [1 , 25].”
Chen, W., Li, C., Chen, D., & Luo, X. (2023) · Neural Networks · doi:10.1016/j.neunet.2023.09.001 · Cited by 23
“Self-supervised learning has proved powerful in learning representations without the need for large labelled datasets [22].”
BackgroundSelf-distillation and self-supervision for partial label learning
Yu, X., Sun, S., & Tian, Y. (2023) · Pattern Recognition · doi:10.1016/j.patcog.2023.110016 · Cited by 22
Self-citationFan, J., Zhou, J., Wang, X., & Wang, J. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3321595 · Cited by 29
“In self-supervised learning, an upstream network is trained with a large amount of unlabeled data, and a downstream network is fine-tuned with a small number of labels [21].”
BackgroundDR-FER: Discriminative and Robust Representation Learning for Facial Expression Recognition
Li, M., Fu, H., He, S., Fan, H., Liu, J., Keppo, J., & Shou, M. Z. (2023) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2023.3347849 · Cited by 32
“Self-supervised visual learning attempts to learn deep representations by exploring intrinsic structures of data rather than human annotated labels [32].”
BackgroundBoundary-aware information maximization for self-supervised medical image segmentation
Peng, J., Wang, P., Pedersoli, M., & Desrosiers, C. (2024) · Medical Image Analysis · doi:10.1016/j.media.2024.103150 · Cited by 24
Few-Shot Font Generation by Learning Style Difference and Similarity
He, X., Zhu, M., Wang, N., & Gao, X. (2024) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2024.3382621 · Cited by 22
“It can be seen as training an encoder for a dictionary look-up task, and it is widely used in the metric learning literature and self-supervised learning (SSL) [53], [54], [55].”
BackgroundAdapting Neural Architectures Between Domains
Li, Y., Yang, Z., Wang, Y., & Xu, C. (2020) · neural information processing systems · Cited by 35
Wu, J. Y., Tamhane, A., Kazanzides, P., & Unberath, M. (2021) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-021-02343-y · Cited by 25
“Examples include video to optical flow transformation, ego-motion pose regression, and audio-visual correspondence [10].”
BackgroundWang, Z., Jiang, F., Liu, T., Xie, F., & Li, P. (2021) · Remote Sensing · doi:10.3390/rs13234927 · Cited by 24
Contrastive Graph Representations for Logical Formulas Embedding
Lin, Q., Liu, J., Zhang, L., Pan, Y., Hu, X., Xu, F., & Zeng, H. (2021) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2021.3139333 · Cited by 31
“Contrastive learning aims to learn high-quality representation via a self-supervised manner, which has gradually emerged its superiority in deep learning [25], [26].”
BackgroundSelf-supervised Modal and View Invariant Feature LearningPreprint
Jing, L., Chen, Y., Zhang, L., He, M., & Tian, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.14169 · Cited by 27
“Since it can be easily scaled up to large-datasets, recently some self-supervised methods achieved comparable or even better performance on some downstream tasks than supervised learning methods [30, 17, 7, 2, 24].”
Self-citationRanking and Tuning Pre-trained Models: A New Paradigm for Exploiting Model HubsPreprint
You, K., Liu, Y., Zhang, Z., Wang, J., Jordan, M. I., & Long, M. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.10545 · Cited by 32
A Closer Look at Benchmarking Self-supervised Pre-training with Image Classification
Marks, M., Knott, M., Kondapaneni, N., Cole, E., Defraeye, T., Pérez‐Cruz, F., & Perona, P. (2025) · International Journal of Computer Vision · doi:10.1007/s11263-025-02402-w · Cited by 37
“…pre-training (Shi et al., 2022), as the latter is prone to overfit on features that are only useful for solving the initial supervised task (Jing & Tian, 2019). c. Semi-supervised Learning 1 : A model is pre-trained on an unlabelled dataset A via SSL, followed by supervised fine-tuning on a…”
Multi-Class Intrusion Detection Using Two-Channel Color Mapping in IEEE 802.11 Wireless Network
Aminanto, M. E., Wicaksono, R. S. H., Aminanto, A. E., Tanuwidjaja, H. C., Yola, L., & Kim, K. (2022) · IEEE Access · doi:10.1109/access.2022.3164104 · Cited by 24
“[12] in 2016, is the most frequently used architecture as a base network [19].”
Classification of Ground-Based Cloud Images by Contrastive Self-Supervised Learning
Lv, Q., Li, Q., Chen, K., Lu, Y., & Wang, L. (2022) · Remote Sensing · doi:10.3390/rs14225821 · Cited by 23
Liu, W., Zhang, H., Chang, S., Wang, H., He, J. L., & Huang, Q. (2022) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2022.106390 · Cited by 21
Gazis, A., Karaiskos, P., & Loukas, C. (2022) · Bioengineering · doi:10.3390/bioengineering9120737 · Cited by 22
“The self-supervision procedure is based on the idea of network training on a pretext task [23], defined as a self-supervised learning problem where the required labels are automatically generated from the input data.”
MethodDense lead contrast for self-supervised representation learning of multilead electrocardiograms
Liu, W., Li, Z., Zhang, H., Chang, S., Wang, H., He, J. L., & Huang, Q. (2023) · Information Sciences · doi:10.1016/j.ins.2023.03.099 · Cited by 24
Benchmarking Polyp Segmentation Methods in Narrow-Band Imaging Colonoscopy Images
Yue, G., Zhuo, G., Li, S., Zhou, T., Du, J., Yan, W., Hou, J., Liu, W., & Wang, T. (2023) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2023.3270724 · Cited by 27
“Recently,techniquesofsemi-supervised[69] andself-supervisedlearning[70]haveshownpowerfulpotential forlearningmorerobustanddiscriminativerepresentationsby leveraging few labeled data and many unlabeled data simultaneously.”
BackgroundTwo-Stage Self-Supervised Cycle-Consistency Transformer Network for Reducing Slice Gap in MR Images
Lu, Z., Wang, J., Li, Z., Ying, S., Wang, J., Shi, J., & Shen, D. G. (2023) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2023.3271815 · Cited by 22
“Deeplearning(DL)hasachievedgreatreputationonthetask ofMRIsuper-resolution(SR)inrecentyears[4],[5],[6],[7],[8], [9],[10],[11],[12],[13],[14],[15],[16],[17],[18].”
MethodArtificial Intelligence and Pathomics
Moghadam, P. A., Bashashati, A., & Goldenberg, S. L. (2023) · Urologic Clinics of North America · doi:10.1016/j.ucl.2023.06.001 · Cited by 27
An Investigation Into the Advancements of Edge-AI Capabilities for Structural Health Monitoring
Mishra, A., Gangisetti, G., & Khazanchi, D. (2024) · IEEE Access · doi:10.1109/access.2024.3365550 · Cited by 21
“Self-supervised learning(SSL) enables models to learn from unlabeled data through tasks that do not require human labeling [58].”
BackgroundMeta-Learning for Wireless Communications: A Survey and a Comparison to GNNs
Zhao, B., Wu, J., Ma, Y., & Yang, C. (2024) · IEEE Open Journal of the Communications Society · doi:10.1109/ojcoms.2024.3380512 · Cited by 31
“…learning: This approach addresses the expensive data collection and manual annotation issue faced by supervised learning [45], [46], [47], [48], which employs a DNN to predict one part of the input data – or an automatically generated pseudo label – given another part of the input…”
MethodTu, L., Li, J., Huang, X., Gong, J., Xie, X., & Wang, L. (2024) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2024.3392962 · Cited by 24
“Through generating supervised signals from data itself to guide the network training [16], SSL can learn abundant representations by leveraging the large amount of unlabeled data [17].”
BackgroundSelf-supervised approach for diabetic retinopathy severity detection using vision transformer
Ohri, K., Kumar, M., & Sukheja, D. (2024) · Progress in Artificial Intelligence · doi:10.1007/s13748-024-00325-0 · Cited by 23
“One of the reasons for the success of self-supervised methods is the reduced cost of annotations that is achieved by utilizing unlabeled medical images [25].”
Huang, Z., Jiang, R., Aeron, S., & Hughes, M. C. (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/cvpr52733.2024.02103 · Cited by 36
“The TissueMNIST hyperparameters originate from our experiments as depicted in Semi-supervised learning trains on the labeled and unlabeled data simultaneously, usually with the total loss being a weighted sum of a labeled loss term and an unlabeled loss term.”
MethodContrastive self-supervised learning for neurodegenerative disorder classification
Gryshchuk, V., Singh, D. S., Teipel, S. J., Dyrba, M., & ADNI, A. (2025) · Frontiers in Neuroinformatics · doi:10.3389/fninf.2025.1527582 · Cited by 24
“Models trained under the SSL approach have found application in different domains, that is, image processing (Jing and Tian, 2020), video processing (Schiappa et al., 2023), and audio processing (Liu et al., 2022a).”
Road Object Detection using SSD-MobileNet Algorithm: Case Study for Real-Time ADAS Applications
Bouazizi, O., Azroumahli, C., Mourabit, A. E., & Oussouaddi, M. (2024) · Journal of Robotics and Control (JRC) · doi:10.18196/jrc.v5i2.21145 · Cited by 21
Exploring DeshuffleGANs in Self-Supervised Generative Adversarial Networks
Baykal, G., Özçelik, F., & Ünal, G. (2021) · Pattern Recognition · doi:10.1016/j.patcog.2021.108244 · Cited by 24
Triantafyllopoulos, A., & Schuller, B. W. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414896 · Cited by 32
Self-Supervised Features Improve Open-World LearningPreprint
Dhamija, A. R., Ahmad, T., Schwan, J., Jafarzadeh, M., Li, C., & Boult, T. E. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.07848 · Cited by 23
“com/Vastlab/SSFiOWL ignored the importance of feature representations, which has been studied in broader settings [31, 44, 19, 33].”
BackgroundSCALa: A blueprint for computational models of language acquisition in social contextPreprint
Tsuji, S., Cristià, A., & Dupoux, E. (2021) · PsyArXiv (OSF Preprints) · doi:10.1016/j.cognition.2021.104779 · Cited by 24
Emam, Z., Kondrich, A., Harrison, S., Lau, F., Wang, Y., Kim, A., & Branson, E. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.00114 · Cited by 27
“(Hendrycks et al., 2019; Jing & Tian, 2019) showed that while self-supervision does not substantially improve accuracy when used in tandem with standard training on annotated datasets, it can improve different aspects of model robustness: noisy labels (Zhang & Sabuncu, 2018), adversarial examples (…”
BackgroundSmall-sample cucumber disease identification based on multimodal self-supervised learning
Cao, Y., Sun, G., Yuan, Y., & Chen, L. (2024) · Crop Protection · doi:10.1016/j.cropro.2024.107006 · Cited by 23
You Only Need Clear Images: Self-Supervised Single Image Dehazing
Chen, J., Ren, W., Zhao, H., Xia, Q., & Yang, G. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2025.3542999 · Cited by 20
Pérez‐García, F., Rodionov, R., Alim‐Marvasti, A., Sparks, R. E., Duncan, J. S., & Ourselin, S. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-59716-0_12 · Cited by 23
Multiview Deep Anomaly Detection: A Systematic Exploration
Wang, S., Liu, J., Yu, G. W., Liu, X., Zhou, S., Zhu, E., Yang, Y., Yin, J., & Yang, W. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3184723 · Cited by 28
“Self-supervised learning is a hot topic in recent research, and it has been demonstrated as a highly effective way to conduct unsupervised representation learning [54].”
BackgroundSelf-Supervised Point Cloud Registration With Deep Versatile Descriptors for Intelligent Driving
Liu, D., Chen, C., Xu, C., Qiu, R. C., & Chu, L. (2023) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3268273 · Cited by 26
“the success of self-supervised learning [51], [52], we design a pretext task to extract informative global and local descriptors.”
MethodSelf-Supervised On-Device Federated Learning From Unlabeled Streams
Shi, J., Wu, Y., Zeng, D., Tao, J., Hu, J., & Shi, Y. (2023) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2023.3274956 · Cited by 23
“Self-Supervised Visual Representation Learning: Self-supervised learning tries to learn effective visual representations from unlabeled data [3], [4], [5], [6], [8], [9].”
BackgroundMa, S., Tong, L., Zhou, J. W., Yu, J. S., & Xiao, C. (2023) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2023.3297216 · Cited by 22
“In addition to FSL, self-supervised learning (SSL) [43], [44] can improve the representation ability of data for small sample problems.”
BackgroundSketch2Saliency: Learning to Detect Salient Objects from Human Drawings
Bhunia, A. K., Koley, S., Kumar, A., Sain, A., Chowdhury, P., Xiang, T., & Song, Y.-Z. (2023) · Computer Vision and Pattern Recognition · doi:10.1109/cvpr52729.2023.00268 · Cited by 28
“Following the linear evaluation protocol of self-supervised learning [31], we resize the last output feature map to the size of input image using bilinear interpolation, and apply one learnable layer for pixel-wise binary classification.”
MethodFeng, H., Jiang, K. L., Zhao, Y., Al‐Malahi, A., & Tang, B. (2023) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2023.3323443 · Cited by 20
“Self-supervised learning [20] has attracted much research interest in computer vision and natural language processing in recent years.”
BackgroundNEMO: A Database for Emotion Analysis Using Functional Near-Infrared Spectroscopy
Spapé, M., Mäkelä, K., & Ruotsalo, T. (2023) · IEEE Transactions on Affective Computing · doi:10.1109/taffc.2023.3315971 · Cited by 22
“…machine learning methods for improved affective decoding, to study the interrelationships between the imagery and perception tasks, and utilize the images used as stimuli as an additional modality via computer vision or generative models [18], collaborative models [20], or self-supervision [42].”
MethodYang, Z., Du, H., Niyato, D. T., Wang, X., Zhou, Y. F., Feng, L., Zhou, F., Li, W., & Qiu, X. (2025) · IEEE Wireless Communications · doi:10.1109/mwc.002.2400197 · Cited by 20
“This means that a pre-trained model using SSL can fine-tuned on specific tasks with minimal additional labeled data, providing strong different applications and reducing the need extensive task-specific training [8].”
Salas, J., Vidal, F. D. B., & Martinez-Trinidad, F. (2019) · IEEE Latin America Transactions · doi:10.1109/tla.2019.9011537 · Cited by 21
“No primeiro caso, divide-se os dados em conjuntos de treinamento-validac¸˜ao e testes [69].”
Tian, S. F., & Xiong, T. (2020) · International World Wide Web Conference (WWW) · doi:10.1145/3366423.3380166 · Cited by 20
“Representation learning can be broadly divided into two categories: the generative model, which learns a compact representation that predicts pixel-level details [12, 27, 31], and self-supervised model, which predicts the representation of future observations from past observations [23, 25].”
BackgroundSelf-supervised endoscopic image key-points matching
Farhat, M., Chaabouni-Chouayakh, H., & Hamadou, A. B. (2022) · Expert Systems with Applications · doi:10.1016/j.eswa.2022.118696 · Cited by 23
Self-supervised Scale Equivariant Network for Weakly Supervised Semantic SegmentationPreprint
Wang, Y., Zhang, J., Kan, M., Shan, S., & Chen, X. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.03714 · Cited by 21
“The task-related feature representations will be learned through this process (Jing and Tian 2019).”
BackgroundSelf-Supervised Learning for Text Recognition: A Critical Survey
Penarrubia, C., Valero-Mas, J. J., & Calvo-Zaragoza, J. (2025) · International Journal of Computer Vision · doi:10.1007/s11263-025-02487-3 · Cited by 33
“This supervision is done by automatically creating labels for the originally unlabeled data (Jing & Tian, 2020).”
InfluentialKinakh, V., Taran, O., & Voloshynovskiy, S. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00129 · Cited by 27
“We refer the reader to [22] for the details of these methods.”
MethodWang, J., Grant, T., Velipasalar, S., Geng, B., & Hirshfield, L. (2021) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2021.3133871 · Cited by 23
Bartels, J., Tokgöz, K. K., Sihan, A., Fukawa, M., Otsubo, S., Li, C., Rachi, I., Takeda, K.-I., Minati, L., & Ito, H. (2022) · IEEE Access · doi:10.1109/access.2022.3156278 · Cited by 19
“Consequently, we are investigating self-supervised learning [36] that has been used in many applications [37], [38] but remains unexplored for cow behavior.”
BackgroundLi, G., Togo, R., Ogawa, T., & Haseyama, M. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9746540 · Cited by 27
“, texture, position, and color) without manually annotated labels [9].”
BackgroundLiu, B., Gao, K., Yu, A., Ding, L., Qiu, C., & Li, J. (2022) · Remote Sensing · doi:10.3390/rs14174236 · Cited by 19
Xu, F., Yan, Y., Zhu, J., Chen, X., Gao, L., Liu, Y., Shi, W., Lou, Y., Wang, W., Leng, J., et al. (2023) · International Journal of Neural Systems · doi:10.1142/s0129065723500661 · Cited by 20
Guo, Q., Liao, Y., Li, Z., & Liang, S. (2023) · Entropy · doi:10.3390/e25101421 · Cited by 19
Visualization for Trust in Machine Learning Revisited: The State of the Field in 2023
Chatzimparmpas, A., Kucher, K., & Kerren, A. (2024) · IEEE Computer Graphics and Applications · doi:10.1109/mcg.2024.3360881 · Cited by 19
“18,19 Similarly, self-supervision is another area of rapid advancement that requires the visualization community ’ s attention, 20 as well as prompt engineering and model checkers for “ hallucinations ” of large language models (LLMs).”
BackgroundLiu, Z., Wu, J., Shen, Z., Chen, X., Wu, Q., Gui, Z., Senhadji, L., & Shu, H. (2024) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2024.3376404 · Cited by 23
“Inspired by this, video representation learning (VRL) methods [26], [27], [28], [29], which have extensive research in video processing, feature and model generalization enhancement, are a worthwhile direction to explore.”
MethodShi, J., Liu, W., Zhang, H., Li, Z., Chang, S., Wang, H., He, J. W., & Huang, Q. (2024) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.106253 · Cited by 20
Exploring the Landscape of Ubiquitous In-home Health Monitoring: A Comprehensive Survey
Pourpanah, F., & Etemad, A. (2024) · ACM Transactions on Computing for Healthcare · doi:10.1145/3670854 · Cited by 24
“For a review on wearable technologies, please see [94, 184].”
Background
Metadata
- Authors
- Jing, L., & Tian, Y.
- Affiliations
- The Graduate Center, CUNY (US)
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2021-11-01
- Volume
- 43
- Issue
- 11
- Pages
- 4037-4058
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- 1939-3539
- OpenAlex
- W3023371261
- Cited by
- 2,036
- References
- 206
- Citations by year
- 2026: 160 · 2025: 283 · 2024: 354 · 2023: 405 · 2022: 377 · 2021: 305
- FWCI
- 119.51
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Science Foundation (IIS-1400802)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Keywords
- self-supervised learning, visual feature learning, imaging features, video features, deep learning architectures, unlabeled data, audio features, 3D data, benchmark dataset, evaluation metrics
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.