--- title: "U-Net: Convolutional Networks for — Ronneberger 2015" description: "Biomedical image analysis often requires a class label for every pixel, yet annotated training images are scarce." authors: ["Olaf Ronneberger", "Philipp Fischer", "Thomas Brox"] year: 2015 venue: "arXiv:1505.04597" source: "https://arxiv.org/abs/1505.04597" doi: "10.48550/arxiv.1505.04597" arxiv: "1505.04597" licence: "arXiv — non-exclusive license to distribute" licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/" summary_licence: "Elamur Cosmic License" canonical: "https://research.elamur.ai/free/en/papers/ronneberger-2015-u-net-convolutional-networks-biomedical/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/ronneberger-2015-u-net-convolutional-networks-biomedical/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2015](https://md.research.elamur.ai/free/en/papers/2015/page/2/) / U-Net: Convolutional Networks for Biomedical Image Segmentation # U-Net: Convolutional Networks for Biomedical Image Segmentation **Authors:** [Olaf Ronneberger](https://md.research.elamur.ai/free/en/authors/olaf-ronneberger/), Philipp Fischer, Thomas Brox · **Year:** 2015 · **Venue:** arXiv:1505.04597 · **Licence:** [arXiv — non-exclusive license to distribute](http://arxiv.org/licenses/nonexclusive-distrib/1.0/) **Links:** [Publisher's page](https://arxiv.org/abs/1505.04597) · [PDF](https://arxiv.org/pdf/1505.04597) · [DOI 10.48550/arxiv.1505.04597](https://doi.org/10.48550/arxiv.1505.04597) · [arXiv 1505.04597](https://arxiv.org/abs/1505.04597) · [OpenAlex](https://openalex.org/W2952232639) **Citations:** 98 citations (openalex-snapshot · 2026-10-14) · [714](#cited-by) in our library, of the 140673 papers whose bibliographies we read ## Summary Biomedical image analysis often requires a class label for every pixel, yet annotated training images are scarce. Earlier sliding-window CNNs were slow and had to trade localization precision against context. The authors build on fully convolutional networks and propose U-Net, a symmetric encoder–decoder. A downsampling branch gathers context. An upsampling branch with many feature channels restores resolution. At each level, cropped high-resolution features from the downsampling side are concatenated into the upsampling side. The model uses only valid convolutions and has no fully connected layers, so arbitrarily large images can be processed tile by tile, with mirrored borders. To cope with tiny datasets, training relies heavily on random elastic deformations. A per-pixel weighted loss emphasizes the thin background gaps between touching cells. Trained end-to-end in Caffe from only a few dozen images, U-Net set a new best warping error on the ISBI EM segmentation challenge. It also clearly won two ISBI 2015 cell tracking datasets. The work showed that accurate dense segmentation is feasible with very little labeled data, and the architecture became a widely reused template. ## Key points - **Architecture:** 23 convolutional layers built from unpadded 3x3 convolutions with ReLU, 2x2 max pooling, and 2x2 up-convolutions, plus a final 1x1 classifier. Weights use Gaussian (He-style) initialization. Training uses SGD with batch size 1 and momentum 0.99. - **Augmentation and loss:** Elastic deformations are generated from random displacements on a coarse 3x3 grid (std 10 px) with bicubic interpolation. The loss weight map uses w0 = 10 and σ ≈ 5 px. - **ISBI EM challenge (30 training images, 512x512, Drosophila VNC):** warping error 0.000353 and Rand error 0.0382. The previous IDSIA sliding-window CNN scored 0.000420 and 0.0504. - **ISBI cell tracking challenge 2015 (IOU):** PhC-U373 (35 images) reached 92% versus 83% for the runner-up. DIC-HeLa (20 images) reached 77.5% versus 46%. - **Efficiency:** A 512x512 image is segmented in under a second on a GPU. Training takes about 10 hours on an NVidia Titan (6 GB). The Caffe code and trained models were released. *Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.* Fields: [Other](https://md.research.elamur.ai/free/en/shelves/other/) ## Cited by Cited by 714 papers of our library cite it (of the 140673 whose bibliographies we read); 7 of them have a page here, 707 not yet. - [Structured Denoising Diffusion Models in Discrete State-Spaces](https://md.research.elamur.ai/free/en/papers/austin-2021-structured-denoising-diffusion-models-discrete/) - [Condenser: a Pre-training Architecture for Dense Retrieval](https://md.research.elamur.ai/free/en/papers/gao-2021-condenser-pre-training-architecture-dense/) - [COMICORDA: Dialogue Act Recognition in Comic Books](https://md.research.elamur.ai/free/en/papers/martinek-2024-comicorda-dialogue-act-recognition-comic/) - [Unifying Large Language Models and Knowledge Graphs: A Roadmap](https://md.research.elamur.ai/free/en/papers/pan-2023-unifying-large-language-models-knowledge/) - [SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis](https://md.research.elamur.ai/free/en/papers/podell-2023-sdxl-improving-latent-diffusion-models/) - [Generative Modeling by Estimating Gradients of the Data Distribution](https://md.research.elamur.ai/free/en/papers/song-2019-generative-modeling-estimating-gradients-data/) - [Denoising Diffusion Implicit Models](https://md.research.elamur.ai/free/en/papers/song-2020-denoising-diffusion-implicit-models/) ## Cite The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @misc{ronneberger2015u, title = {{U-Net: Convolutional Networks for Biomedical Image Segmentation}}, author = {Ronneberger, Olaf and Fischer, Philipp and Brox, Thomas}, year = {2015}, doi = {10.48550/arxiv.1505.04597}, eprint = {1505.04597}, archivePrefix = {arXiv}, url = {https://doi.org/10.48550/arxiv.1505.04597} } ``` - [← previous Two/Too Simple Adaptations of Word2Vec for Syntax Problems](https://md.research.elamur.ai/free/en/papers/ling-2015-two-too-simple-adaptations-word2vec/) - [next → Universal Dependencies for Croatian (that work for Serbian, too)](https://md.research.elamur.ai/free/en/papers/agic-2015-universal-dependencies-croatian-that-work/) - [↑ back to papers of 2015](https://md.research.elamur.ai/free/en/papers/2015/page/2/) ## References 14 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Ciresan, D.C., Gambardella, L.M., Giusti, A., Schmidhuber, J.: Deep neural networks segment neuronal membranes in electron microscopy images. In: NIPS. pp. 2852–2860 (2012) [find](https://scholar.google.com/scholar?q=Ciresan%2C+D%2EC%2E%2C+Gambardella%2C+L%2EM%2E%2C+Giusti%2C+A%2E%2C+Schmidhuber%2C+J%2E%3A+Deep+neural+networks+segment+neuronal+membranes+in+electron+microscopy+images%2E+In%3A+NIPS%2E+pp%2E+2852%E2%80%932860+%282012%29) 2. Dosovitskiy, A., Springenberg, J.T., Riedmiller, M., Brox, T.: Discriminative unsupervised feature learning with convolutional neural networks. In: NIPS (2014) [find](https://scholar.google.com/scholar?q=Dosovitskiy%2C+A%2E%2C+Springenberg%2C+J%2ET%2E%2C+Riedmiller%2C+M%2E%2C+Brox%2C+T%2E%3A+Discriminative+unsupervised+feature+learning+with+convolutional+neural+networks%2E+In%3A+NIPS+%282014%29) 3. Girshick, R., Donahue, J., Darrell, T., Malik, J.: Rich feature hierarchies for accurate object detection and semantic segmentation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2014) [find](https://scholar.google.com/scholar?q=Girshick%2C+R%2E%2C+Donahue%2C+J%2E%2C+Darrell%2C+T%2E%2C+Malik%2C+J%2E%3A+Rich+feature+hierarchies+for+accurate+object+detection+and+semantic+segmentation%2E+In%3A+Proceedings+of+the+IEEE+Conference+on+Computer+Vision+and+Patte) [our card](https://md.research.elamur.ai/free/en/papers/girshick-2013-rich-feature-hierarchies-accurate-object/) 4. Hariharan, B., Arbelez, P., Girshick, R., Malik, J.: Hypercolumns for object segmentation and fine-grained localization (2014), arXiv:1411.5752 [cs.CV] — [arXiv:1411.5752](https://arxiv.org/abs/1411.5752) [find](https://scholar.google.com/scholar?q=Hariharan%2C+B%2E%2C+Arbelez%2C+P%2E%2C+Girshick%2C+R%2E%2C+Malik%2C+J%2E%3A+Hypercolumns+for+object+segmentation+and+fine%2Dgrained+localization+%282014%29%2C+arXiv%3A1411%2E5752+%5Bcs%2ECV%5D+%E2%80%94+%5BarXiv%3A1411%2E5752%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F1411%2E5) 5. He, K., Zhang, X., Ren, S., Sun, J.: Delving deep into rectifiers: Surpassing human-level performance on imagenet classification (2015), arXiv:1502.01852 [cs.CV] [find](https://scholar.google.com/scholar?q=He%2C+K%2E%2C+Zhang%2C+X%2E%2C+Ren%2C+S%2E%2C+Sun%2C+J%2E%3A+Delving+deep+into+rectifiers%3A+Surpassing+human%2Dlevel+performance+on+imagenet+classification+%282015%29%2C+arXiv%3A1502%2E01852+%5Bcs%2ECV%5D) 6. Jia, Y., Shelhamer, E., Donahue, J., Karayev, S., Long, J., Girshick, R., Guadarrama, S., Darrell, T.: Caffe: Convolutional architecture for fast feature embedding (2014), arXiv:1408.5093 [cs.CV] [find](https://scholar.google.com/scholar?q=Jia%2C+Y%2E%2C+Shelhamer%2C+E%2E%2C+Donahue%2C+J%2E%2C+Karayev%2C+S%2E%2C+Long%2C+J%2E%2C+Girshick%2C+R%2E%2C+Guadarrama%2C+S%2E%2C+Darrell%2C+T%2E%3A+Caffe%3A+Convolutional+architecture+for+fast+feature+embedding+%282014%29%2C+arXiv%3A1408%2E5093+%5Bcs%2ECV%5D) [our card](https://md.research.elamur.ai/free/en/papers/jia-2014-caffe-convolutional-architecture-fast-feature/) 7. Krizhevsky, A., Sutskever, I., Hinton, G.E.: Imagenet classification with deep convolutional neural networks. In: NIPS. pp. 1106–1114 (2012) [find](https://scholar.google.com/scholar?q=Krizhevsky%2C+A%2E%2C+Sutskever%2C+I%2E%2C+Hinton%2C+G%2EE%2E%3A+Imagenet+classification+with+deep+convolutional+neural+networks%2E+In%3A+NIPS%2E+pp%2E+1106%E2%80%931114+%282012%29) 8. LeCun, Y., Boser, B., Denker, J.S., Henderson, D., Howard, R.E., Hubbard, W., Jackel, L.D.: Backpropagation applied to handwritten zip code recognition. Neural Computation 1(4), 541–551 (1989) [find](https://scholar.google.com/scholar?q=LeCun%2C+Y%2E%2C+Boser%2C+B%2E%2C+Denker%2C+J%2ES%2E%2C+Henderson%2C+D%2E%2C+Howard%2C+R%2EE%2E%2C+Hubbard%2C+W%2E%2C+Jackel%2C+L%2ED%2E%3A+Backpropagation+applied+to+handwritten+zip+code+recognition%2E+Neural+Computation+1%284%29%2C+541%E2%80%93551+%281989%29) 9. Long, J., Shelhamer, E., Darrell, T.: Fully convolutional networks for semantic segmentation (2014), arXiv:1411.4038 [cs.CV] — [arXiv:1411.4038](https://arxiv.org/abs/1411.4038) [find](https://scholar.google.com/scholar?q=Long%2C+J%2E%2C+Shelhamer%2C+E%2E%2C+Darrell%2C+T%2E%3A+Fully+convolutional+networks+for+semantic+segmentation+%282014%29%2C+arXiv%3A1411%2E4038+%5Bcs%2ECV%5D+%E2%80%94+%5BarXiv%3A1411%2E4038%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F1411%2E4038%29) 10. Maska, M., (...), de Solorzano, C.O.: A benchmark for comparison of cell tracking algorithms. Bioinformatics 30, 1609–1617 (2014) [find](https://scholar.google.com/scholar?q=Maska%2C+M%2E%2C+%28%2E%2E%2E%29%2C+de+Solorzano%2C+C%2EO%2E%3A+A+benchmark+for+comparison+of+cell+tracking+algorithms%2E+Bioinformatics+30%2C+1609%E2%80%931617+%282014%29) 11. Seyedhosseini, M., Sajjadi, M., Tasdizen, T.: Image segmentation with cascaded hierarchical models and logistic disjunctive normal networks. In: Computer Vision (ICCV), 2013 IEEE International Conference on. pp. 2168–2175 (2013) [find](https://scholar.google.com/scholar?q=Seyedhosseini%2C+M%2E%2C+Sajjadi%2C+M%2E%2C+Tasdizen%2C+T%2E%3A+Image+segmentation+with+cascaded+hierarchical+models+and+logistic+disjunctive+normal+networks%2E+In%3A+Computer+Vision+%28ICCV%29%2C+2013+IEEE+International+Confere) 12. Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition (2014), arXiv:1409.1556 [cs.CV] [find](https://scholar.google.com/scholar?q=Simonyan%2C+K%2E%2C+Zisserman%2C+A%2E%3A+Very+deep+convolutional+networks+for+large%2Dscale+image+recognition+%282014%29%2C+arXiv%3A1409%2E1556+%5Bcs%2ECV%5D) [our card](https://md.research.elamur.ai/free/en/papers/simonyan-2014-very-deep-convolutional-networks-large/) 13. WWW: Web page of the cell tracking challenge, [http://www.codesolorzano.com/celltrackingchallenge/Cell_Tracking_Challenge/Welcome.html](http://www.codesolorzano.com/celltrackingchallenge/Cell_Tracking_Challenge/Welcome.html) [↗](http://www.codesolorzano.com/celltrackingchallenge/Cell_Tracking_Challenge/Welcome.html) 14. WWW: Web page of the em segmentation challenge, [http://brainiac2.mit.edu/isbi_challenge/U-net](http://brainiac2.mit.edu/isbi_challenge/U-net) implementation, trained networks and supplementary material available at http://lmb.informatik.uni-freiburg.de/people/ronneber/u-net [↗](http://brainiac2.mit.edu/isbi_challenge/U-net)