{"url":"/task/self-supervised-image-classification","name":"Self-Supervised Image Classification","slug":"self-supervised-image-classification","description_markdown":"This is the task of image classification using representations learnt with self-supervised learning. Self-supervised methods generally involve a pretext task that is solved to learn a good representation and a loss function to learn with. One example of a loss function is an autoencoder based loss where the goal is reconstruction of an image pixel-by-pixel. A more popular recent example is a contrastive loss, which measure the similarity of sample pairs in a representation space, and where there can be a varying target instead of a fixed target to reconstruct (as in the case of autoencoders).\r\n\r\nA common evaluation protocol is to train a linear classifier on top of (frozen) representations learnt by self-supervised methods. The leaderboards for the linear evaluation protocol can be found below. In practice, it is more common to fine-tune features on a downstream task. An alternative evaluation protocol therefore uses semi-supervised learning and finetunes on a % of the labels. The leaderboards for the finetuning protocol can be accessed [here](https://paperswithcode.com/task/semi-supervised-image-classification).\r\n\r\nYou may want to read some blog posts before reading the papers and checking the leaderboards:\r\n\r\n- [Contrastive Self-Supervised Learning](https://ankeshanand.com/blog/2020/01/26/contrative-self-supervised-learning.html) - Ankesh Anand\r\n- [The Illustrated Self-Supervised Learning](https://amitness.com/2020/02/illustrated-self-supervised-learning/) - Amit Chaudhary\r\n- [Self-supervised learning and computer vision](https://www.fast.ai/2020/01/13/self_supervised/) - Jeremy Howard\r\n- [Self-Supervised Representation Learning](https://lilianweng.github.io/lil-log/2019/11/10/self-supervised-learning.html) - Lilian Weng\r\n\r\nThere is also Yann LeCun's talk at AAAI-20 which you can watch [here](https://vimeo.com/390347111) (35:00+).\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [A Simple Framework for Contrastive Learning of Visual Representations](https://arxiv.org/pdf/2002.05709v1.pdf) )</span>","categories":[{"name":"Adversarial","url":"/area/adversarial"},{"name":"Audio","url":"/area/audio"},{"name":"Computer Code","url":"/area/computer-code"},{"name":"Computer Vision","url":"/area/computer-vision"},{"name":"Medical","url":"/area/medical"},{"name":"Methodology","url":"/area/methodology"},{"name":"Miscellaneous","url":"/area/miscellaneous"},{"name":"Music","url":"/area/music"},{"name":"Natural Language Processing","url":"/area/natural-language-processing"},{"name":"Reasoning","url":"/area/reasoning"},{"name":"Speech","url":"/area/speech"}],"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","slug_source":"archive_url"},"counts":{"papers_tagged":110,"papers_with_code":97,"benchmarks":3,"benchmark_tables_in_archive":3,"benchmark_tables_shown":3,"benchmark_tables_withheld_as_spam":0,"benchmark_definition":"a leaderboard table with at least one row; benchmark_tables_shown also counts the zero-row tables; benchmark_tables_in_archive adds the tables withheld as spam","datasets":3,"subtasks":0,"parent_tasks":1},"benchmarks":[{"leaderboard":"/sota/self-supervised-image-classification-on","slug":"self-supervised-image-classification-on","dataset":"ImageNet","dataset_url":"/dataset/imagenet","rows_in_archive":144,"metrics":["Top 1 Accuracy","Top 5 Accuracy","Number of Params"],"first_row_in_archive_order":{"model":"DINOv2+reg (ViT-g/14)","paper_title":"Vision Transformers Need Registers","paper_url":"/paper/vision-transformers-need-registers","paper_date":"2023-09-28","arxiv_id":"2309.16588","code_links":[{"title":"rwightman/pytorch-image-models","url":"https://github.com/rwightman/pytorch-image-models"},{"title":"facebookresearch/dinov2","url":"https://github.com/facebookresearch/dinov2"},{"title":"locuslab/massive-activations","url":"https://github.com/locuslab/massive-activations"},{"title":"borisdayma/clip-jax","url":"https://github.com/borisdayma/clip-jax"},{"title":"nickjiang2378/test-time-registers","url":"https://github.com/nickjiang2378/test-time-registers"},{"title":"birder/birder","url":"https://gitlab.com/birder/birder"}],"syntology":{"n":20,"n_ran":4,"n_unverified":16,"n_pointer_only":2}}},{"leaderboard":"/sota/self-supervised-image-classification-on-1","slug":"self-supervised-image-classification-on-1","dataset":"ImageNet (finetuned)","dataset_url":"/dataset/imagenet","rows_in_archive":65,"metrics":["Top 1 Accuracy","Number of Params"],"first_row_in_archive_order":{"model":"DINOv2 (ViT-g/14, 448)","paper_title":"DINOv2: Learning Robust Visual Features without Supervision","paper_url":"/paper/dinov2-learning-robust-visual-features","paper_date":"2023-04-14","arxiv_id":"2304.07193","code_links":[{"title":"huggingface/transformers","url":"https://github.com/huggingface/transformers"},{"title":"facebookresearch/dinov2","url":"https://github.com/facebookresearch/dinov2"},{"title":"roboflow/rf-detr","url":"https://github.com/roboflow/rf-detr"},{"title":"open-edge-platform/training_extensions","url":"https://github.com/open-edge-platform/training_extensions"},{"title":"OML-Team/open-metric-learning","url":"https://github.com/OML-Team/open-metric-learning"},{"title":"leondgarse/keras_cv_attention_models","url":"https://github.com/leondgarse/keras_cv_attention_models/tree/main/keras_cv_attention_models/beit"},{"title":"fabio-sim/Depth-Anything-ONNX","url":"https://github.com/fabio-sim/Depth-Anything-ONNX"},{"title":"open-edge-platform/geti","url":"https://github.com/open-edge-platform/geti"},{"title":"facebookresearch/highrescanopyheight","url":"https://github.com/facebookresearch/highrescanopyheight"},{"title":"PaddlePaddle/PASSL","url":"https://github.com/PaddlePaddle/PASSL"},{"title":"beneroth13/dinov2","url":"https://github.com/beneroth13/dinov2"},{"title":"mohammedsb/dinov2formedical","url":"https://github.com/mohammedsb/dinov2formedical"},{"title":"marrlab/dinobloom","url":"https://github.com/marrlab/dinobloom"},{"title":"bespontaneous/proteus-pytorch","url":"https://github.com/bespontaneous/proteus-pytorch"},{"title":"ByungKwanLee/Causal-Unsupervised-Segmentation","url":"https://github.com/ByungKwanLee/Causal-Unsupervised-Segmentation"},{"title":"zhu-xlab/softcon","url":"https://github.com/zhu-xlab/softcon"},{"title":"birder/birder","url":"https://gitlab.com/birder/birder"},{"title":"gorkaydemir/DINOSAUR","url":"https://github.com/gorkaydemir/DINOSAUR"},{"title":"seatizendoi/dinovdeau","url":"https://github.com/seatizendoi/dinovdeau"},{"title":"BurguerJohn/global_perceptual_similarity_loss","url":"https://github.com/BurguerJohn/global_perceptual_similarity_loss"},{"title":"buyeah1109/KEN","url":"https://github.com/buyeah1109/KEN"},{"title":"JHKim-snu/PGA","url":"https://github.com/JHKim-snu/PGA"},{"title":"BurguerJohn/torch-felix","url":"https://github.com/BurguerJohn/torch-felix"},{"title":"2024-MindSpore-1/Code2","url":"https://github.com/2024-MindSpore-1/Code2/tree/main/model-1/dinov2"},{"title":"buyeah1109/finc","url":"https://github.com/buyeah1109/finc"},{"title":"pwc-1/Paper-8","url":"https://github.com/pwc-1/Paper-8/tree/main/dinov2"}],"syntology":{"n":46,"n_ran":21,"n_unverified":25,"n_pointer_only":12}}},{"leaderboard":"/sota/self-supervised-image-classification-on-chest","slug":"self-supervised-image-classification-on-chest","dataset":"Chest X-ray images","dataset_url":"/dataset/chest-x-ray-images","rows_in_archive":1,"metrics":["Accuracy"],"first_row_in_archive_order":{"model":"DINO-CXR","paper_title":"DINO-CXR: A self supervised method based on vision transformer for chest X-ray classification","paper_url":"/paper/dino-cxr-a-self-supervised-method-based-on","paper_date":"2023-08-01","arxiv_id":"2308.00475","code_links":[],"syntology":null}}],"datasets":[{"url":"/dataset/imagenet","name":"ImageNet","full_name":"","num_papers_in_archive":15430},{"url":"/dataset/chest-x-ray-images","name":"Chest X-ray images","full_name":"chest X-ray images for pneumonia detection","num_papers_in_archive":9},{"url":"/dataset/bioscan-5m","name":"BIOSCAN-5M","full_name":"","num_papers_in_archive":4}],"subtasks":[],"parent_tasks":[{"url":"/task/image-classification","name":"Image Classification"}],"papers":{"order":"repositories listed in the archive (desc), then date (desc); the archive holds no stars","population":"papers tagged with this task that list at least one repository in the archive","shown":30,"of":97,"tagged_in_all":110,"items":[{"url":"/paper/a-simple-framework-for-contrastive-learning","title":"A Simple Framework for Contrastive Learning of Visual Representations","date":"2020-02-13","arxiv_id":"2002.05709","repositories_listed":96,"syntology":{"n":137,"n_ran":79,"n_unverified":58,"n_pointer_only":52}},{"url":"/paper/masked-autoencoders-are-scalable-vision","title":"Masked Autoencoders Are Scalable Vision Learners","date":"2021-11-11","arxiv_id":"2111.06377","repositories_listed":58,"syntology":{"n":137,"n_ran":71,"n_unverified":66,"n_pointer_only":73}},{"url":"/paper/momentum-contrast-for-unsupervised-visual","title":"Momentum Contrast for Unsupervised Visual Representation Learning","date":"2019-11-13","arxiv_id":"1911.05722","repositories_listed":44,"syntology":{"n":42,"n_ran":26,"n_unverified":16,"n_pointer_only":16}},{"url":"/paper/colorful-image-colorization","title":"Colorful Image Colorization","date":"2016-03-28","arxiv_id":"1603.08511","repositories_listed":39,"syntology":{"n":73,"n_ran":32,"n_unverified":41,"n_pointer_only":41}},{"url":"/paper/improved-baselines-with-momentum-contrastive","title":"Improved Baselines with Momentum Contrastive Learning","date":"2020-03-09","arxiv_id":"2003.04297","repositories_listed":36,"syntology":{"n":43,"n_ran":8,"n_unverified":35,"n_pointer_only":10}},{"url":"/paper/emerging-properties-in-self-supervised-vision","title":"Emerging Properties in Self-Supervised Vision Transformers","date":"2021-04-29","arxiv_id":"2104.14294","repositories_listed":32,"syntology":{"n":20,"n_ran":5,"n_unverified":15,"n_pointer_only":2}},{"url":"/paper/bootstrap-your-own-latent-a-new-approach-to","title":"Bootstrap your own latent: A new approach to self-supervised Learning","date":"2020-06-13","arxiv_id":"2006.07733","repositories_listed":31,"syntology":{"n":79,"n_ran":62,"n_unverified":17,"n_pointer_only":46}},{"url":"/paper/representation-learning-with-contrastive","title":"Representation Learning with Contrastive Predictive Coding","date":"2018-07-10","arxiv_id":"1807.03748","repositories_listed":28,"syntology":{"n":45,"n_ran":29,"n_unverified":16,"n_pointer_only":22}},{"url":"/paper/dinov2-learning-robust-visual-features","title":"DINOv2: Learning Robust Visual Features without Supervision","date":"2023-04-14","arxiv_id":"2304.07193","repositories_listed":26,"syntology":{"n":46,"n_ran":21,"n_unverified":25,"n_pointer_only":12}},{"url":"/paper/exploring-simple-siamese-representation","title":"Exploring Simple Siamese Representation Learning","date":"2020-11-20","arxiv_id":"2011.10566","repositories_listed":26,"syntology":{"n":39,"n_ran":34,"n_unverified":5,"n_pointer_only":22}},{"url":"/paper/barlow-twins-self-supervised-learning-via","title":"Barlow Twins: Self-Supervised Learning via Redundancy Reduction","date":"2021-03-04","arxiv_id":"2103.03230","repositories_listed":24,"syntology":{"n":26,"n_ran":21,"n_unverified":5,"n_pointer_only":10}},{"url":"/paper/unsupervised-representation-learning-by-1","title":"Unsupervised Representation Learning by Predicting Image Rotations","date":"2018-03-21","arxiv_id":"1803.07728","repositories_listed":20,"syntology":{"n":22,"n_ran":14,"n_unverified":8,"n_pointer_only":18}},{"url":"/paper/resmlp-feedforward-networks-for-image","title":"ResMLP: Feedforward networks for image classification with data-efficient training","date":"2021-05-07","arxiv_id":"2105.03404","repositories_listed":19,"syntology":{"n":7,"n_ran":2,"n_unverified":5,"n_pointer_only":0}},{"url":"/paper/unsupervised-learning-of-visual-features-by","title":"Unsupervised Learning of Visual Features by Contrasting Cluster Assignments","date":"2020-06-17","arxiv_id":"2006.09882","repositories_listed":18,"syntology":{"n":17,"n_ran":13,"n_unverified":4,"n_pointer_only":6}},{"url":"/paper/beit-bert-pre-training-of-image-transformers","title":"BEiT: BERT Pre-Training of Image Transformers","date":"2021-06-15","arxiv_id":"2106.08254","repositories_listed":14,"syntology":{"n":11,"n_ran":6,"n_unverified":5,"n_pointer_only":0}},{"url":"/paper/xcit-cross-covariance-image-transformers","title":"XCiT: Cross-Covariance Image Transformers","date":"2021-06-17","arxiv_id":"2106.09681","repositories_listed":12,"syntology":{"n":14,"n_ran":3,"n_unverified":11,"n_pointer_only":3}},{"url":"/paper/an-empirical-study-of-training-self","title":"An Empirical Study of Training Self-Supervised Vision Transformers","date":"2021-04-05","arxiv_id":"2104.02057","repositories_listed":9,"syntology":{"n":2,"n_ran":2,"n_unverified":0,"n_pointer_only":2}},{"url":"/paper/big-self-supervised-models-are-strong-semi","title":"Big Self-Supervised Models are Strong Semi-Supervised Learners","date":"2020-06-17","arxiv_id":"2006.10029","repositories_listed":9,"syntology":{"n":6,"n_ran":0,"n_unverified":6,"n_pointer_only":0}},{"url":"/paper/deep-clustering-for-unsupervised-learning-of","title":"Deep Clustering for Unsupervised Learning of Visual Features","date":"2018-07-15","arxiv_id":"1807.05520","repositories_listed":9,"syntology":{"n":7,"n_ran":5,"n_unverified":2,"n_pointer_only":4}},{"url":"/paper/contrastive-multiview-coding","title":"Contrastive Multiview Coding","date":"2019-06-13","arxiv_id":"1906.05849","repositories_listed":8,"syntology":{"n":4,"n_ran":3,"n_unverified":1,"n_pointer_only":1}},{"url":"/paper/simmim-a-simple-framework-for-masked-image","title":"SimMIM: A Simple Framework for Masked Image Modeling","date":"2021-11-18","arxiv_id":"2111.09886","repositories_listed":7,"syntology":{"n":14,"n_ran":9,"n_unverified":5,"n_pointer_only":0}},{"url":"/paper/self-supervised-learning-of-pretext-invariant","title":"Self-Supervised Learning of Pretext-Invariant Representations","date":"2019-12-04","arxiv_id":"1912.01991","repositories_listed":7,"syntology":{"n":4,"n_ran":3,"n_unverified":1,"n_pointer_only":4}},{"url":"/paper/vision-transformers-need-registers","title":"Vision Transformers Need Registers","date":"2023-09-28","arxiv_id":"2309.16588","repositories_listed":6,"syntology":{"n":20,"n_ran":4,"n_unverified":16,"n_pointer_only":2}},{"url":"/paper/eva-exploring-the-limits-of-masked-visual","title":"EVA: Exploring the Limits of Masked Visual Representation Learning at Scale","date":"2022-11-14","arxiv_id":"2211.07636","repositories_listed":6,"syntology":{"n":3,"n_ran":1,"n_unverified":2,"n_pointer_only":0}},{"url":"/paper/context-autoencoder-for-self-supervised","title":"Context Autoencoder for Self-Supervised Representation Learning","date":"2022-02-07","arxiv_id":"2202.03026","repositories_listed":6,"syntology":null},{"url":"/paper/masked-feature-prediction-for-self-supervised","title":"Masked Feature Prediction for Self-Supervised Visual Pre-Training","date":"2021-12-16","arxiv_id":"2112.09133","repositories_listed":6,"syntology":null},{"url":"/paper/vicreg-variance-invariance-covariance","title":"VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning","date":"2021-05-11","arxiv_id":"2105.04906","repositories_listed":6,"syntology":{"n":17,"n_ran":11,"n_unverified":6,"n_pointer_only":1}},{"url":"/paper/self-supervised-learning-with-swin","title":"Self-Supervised Learning with Swin Transformers","date":"2021-05-10","arxiv_id":"2105.04553","repositories_listed":6,"syntology":null},{"url":"/paper/revisiting-self-supervised-visual","title":"Revisiting Self-Supervised Visual Representation Learning","date":"2019-01-25","arxiv_id":"1901.09005","repositories_listed":6,"syntology":{"n":15,"n_ran":0,"n_unverified":15,"n_pointer_only":7}},{"url":"/paper/self-labelling-via-simultaneous-clustering-1","title":"Self-labelling via simultaneous clustering and representation learning","date":"2019-11-13","arxiv_id":"1911.05371","repositories_listed":5,"syntology":{"n":18,"n_ran":7,"n_unverified":11,"n_pointer_only":3}}],"syntology_records":27,"syntology_note":"a paper without a record is not a recorded non-run: it may lack an arXiv id or simply be absent from the graph layer"},"description_links":{"kept":1,"unwrapped_to_text":0,"bare_urls_linked":0,"relative_images_dropped":0,"rule":"internal links are kept only when the target slug exists in the catalog"},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per-sample execution status on synthesized fixtures ('ran N of M samples'); not a correctness claim and not a ranking signal.","status_vocabulary":{"ran_honours":"ran, honoured the contract we drafted","ran_violates":"ran, violated the contract we drafted","ran_draft_wrong":"ran; our contract draft was wrong, not the code","ran_fixture":"ran; our fixture could not drive it","ran":"ran on a synthesized input","unverified":"unverified (harvested, no recorded run)"}},"not_shown":{"libraries":"the archive has no per-task library table","trend_sparklines":"the Trend column of the benchmarks table was a rendered image; it is not in the archive","social_and_latest_sorts":"stars and social signals are not in the archive"}}