{"url":"/sota/image-to-image-translation-on-cityscapes","task":{"name":"Image-to-Image Translation","url":"/task/image-to-image-translation","note":null},"dataset":{"name":"Cityscapes Labels-to-Photo","url":"/dataset/cityscapes"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"**Image-to-Image Translation** is a task in computer vision and machine learning where the goal is to learn a mapping between an input image and an output image, such that the output image can be used to perform a specific task, such as style transfer, data augmentation, or image restoration.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [Unpaired Image-to-Image Translation\r\nusing Cycle-Consistent Adversarial Networks](https://arxiv.org/pdf/1703.10593v6.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["mIoU","FID","Accuracy","Class IOU","Per-class Accuracy","Per-pixel Accuracy","LPIPS"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"mIoU":null,"FID":"lower","Accuracy":"higher","Class IOU":"higher","Per-class Accuracy":"higher","Per-pixel Accuracy":"higher","LPIPS":null}},"counts":{"rows":21,"rows_with_code":17,"rows_with_paper_page":21,"rows_dated":21,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"DP-SIMS (ConvNext-L)","metrics":{"FID":"38.2","mIoU":"76.3"},"uses_additional_data":false,"paper_date":"2023-12-20","paper":"/paper/unlocking-pre-trained-image-backbones-for","paper_url":"https://arxiv.org/abs/2312.13314v2","paper_title":"Unlocking Pre-trained Image Backbones for Semantic Image Synthesis","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":2,"model":"DP-GAN","metrics":{"FID":"44.1","mIoU":"73.6"},"uses_additional_data":false,"paper_date":"2022-10-08","paper":"/paper/dual-pyramid-generative-adversarial-networks","paper_url":"https://arxiv.org/abs/2210.04085v1","paper_title":"Dual Pyramid Generative Adversarial Networks for Semantic Image Synthesis","code":"https://github.com/sj-li/dp_gan","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"OASIS","metrics":{"FID":"47.7","LPIPS":"0.275","mIoU":"69.3"},"uses_additional_data":false,"paper_date":"2020-12-08","paper":"/paper/you-only-need-adversarial-supervision-for-1","paper_url":"https://arxiv.org/abs/2012.04781v3","paper_title":"You Only Need Adversarial Supervision for Semantic Image Synthesis","code":"https://github.com/boschresearch/OASIS","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":6,"n_samples":7,"n_pointer_only_licence":7}},{"rank_in_archive_order":4,"model":"SPADE + SESAME","metrics":{"FID":"54.2","Per-pixel Accuracy":"82.5%","mIoU":"66"},"uses_additional_data":false,"paper_date":"2020-04-10","paper":"/paper/sesame-semantic-editing-of-scenes-by-adding","paper_url":"https://arxiv.org/abs/2004.04977v2","paper_title":"SESAME: Semantic Editing of Scenes by Adding, Manipulating or Erasing Objects","code":"https://github.com/vglsd/OpenSESAME","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"CC-FPSE","metrics":{"FID":"54.3","LPIPS":"0.073","Per-pixel Accuracy":"82.3%","mIoU":"65.5"},"uses_additional_data":false,"paper_date":"2019-10-15","paper":"/paper/learning-to-predict-layout-to-image","paper_url":"https://arxiv.org/abs/1910.06809v3","paper_title":"Learning to Predict Layout-to-image Conditional Convolutions for Semantic Image Synthesis","code":"https://github.com/xh-liu/CC-FPSE","n_code_links":1,"syntology":null},{"rank_in_archive_order":6,"model":"SPADE + FFL","metrics":{"FID":"59.5","Per-pixel Accuracy":"82.5%","mIoU":"64.2"},"uses_additional_data":false,"paper_date":"2020-12-23","paper":"/paper/focal-frequency-loss-for-generative-models","paper_url":"https://arxiv.org/abs/2012.12821v3","paper_title":"Focal Frequency Loss for Image Reconstruction and Synthesis","code":"https://github.com/EndlessSora/focal-frequency-loss","n_code_links":1,"syntology":null},{"rank_in_archive_order":7,"model":"CC-FPSE-AUG","metrics":{"Accuracy":"93.5","FID":"52.1","mIoU":"63.1"},"uses_additional_data":false,"paper_date":"2020-11-25","paper":"/paper/improving-augmentation-and-evaluation-schemes","paper_url":"https://arxiv.org/abs/2011.12636v3","paper_title":"Improving Augmentation and Evaluation Schemes for Semantic Image Synthesis","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":8,"model":"SPADE","metrics":{"FID":"71.8","Per-pixel Accuracy":"81.9%","mIoU":"62.3"},"uses_additional_data":false,"paper_date":"2019-03-18","paper":"/paper/semantic-image-synthesis-with-spatially","paper_url":"https://arxiv.org/abs/1903.07291v2","paper_title":"Semantic Image Synthesis with Spatially-Adaptive Normalization","code":"https://github.com/NVlabs/SPADE","n_code_links":24,"syntology":{"n_ran":8,"n_unverified":1,"n_samples":9,"n_pointer_only_licence":2}},{"rank_in_archive_order":9,"model":"pix2pixHD","metrics":{"FID":"95","Per-pixel Accuracy":"81.4%","mIoU":"58.3"},"uses_additional_data":false,"paper_date":"2017-11-30","paper":"/paper/high-resolution-image-synthesis-and-semantic","paper_url":"http://arxiv.org/abs/1711.11585v2","paper_title":"High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs","code":"https://github.com/NVIDIA/pix2pixHD","n_code_links":21,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":3}},{"rank_in_archive_order":10,"model":"Pix2PixHD-AUG","metrics":{"Accuracy":"92.7","FID":"72.7","mIoU":"58"},"uses_additional_data":false,"paper_date":"2020-11-25","paper":"/paper/improving-augmentation-and-evaluation-schemes","paper_url":"https://arxiv.org/abs/2011.12636v3","paper_title":"Improving Augmentation and Evaluation Schemes for Semantic Image Synthesis","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":11,"model":"CRN","metrics":{"FID":"104.7","Per-pixel Accuracy":"77.1%","mIoU":"52.4"},"uses_additional_data":false,"paper_date":"2017-07-28","paper":"/paper/photographic-image-synthesis-with-cascaded","paper_url":"http://arxiv.org/abs/1707.09405v1","paper_title":"Photographic Image Synthesis with Cascaded Refinement Networks","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":12,"model":"SIMS","metrics":{"FID":"49.7","Per-pixel Accuracy":"75.5%","mIoU":"47.2"},"uses_additional_data":false,"paper_date":"2018-04-29","paper":"/paper/semi-parametric-image-synthesis","paper_url":"http://arxiv.org/abs/1804.10992v1","paper_title":"Semi-parametric Image Synthesis","code":"https://github.com/xjqicuhk/SIMS","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"USIS","metrics":{"FID":"53.67","mIoU":"44.78"},"uses_additional_data":false,"paper_date":"2021-09-29","paper":"/paper/usis-unsupervised-semantic-image-synthesis","paper_url":"https://arxiv.org/abs/2109.14715v1","paper_title":"USIS: Unsupervised Semantic Image Synthesis","code":"https://github.com/GeorgeEskandar/USIS-Unsupervised-Semantic-Image-Synthesis","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"USIS-Wavelet","metrics":{"FID":"50.14","mIoU":"42.32"},"uses_additional_data":false,"paper_date":"2023-05-16","paper":"/paper/wavelet-based-unsupervised-label-to-image-1","paper_url":"https://arxiv.org/abs/2305.09647v1","paper_title":"Wavelet-based Unsupervised Label-to-Image Translation","code":"https://github.com/GeorgeEskandar/USIS-Unsupervised-Semantic-Image-Synthesis","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"SB-GAN","metrics":{"FID":"60.39"},"uses_additional_data":false,"paper_date":"2019-11-26","paper":"/paper/semantic-bottleneck-scene-generation","paper_url":"https://arxiv.org/abs/1911.11357v1","paper_title":"Semantic Bottleneck Scene Generation","code":"https://github.com/azadis/SB-GAN","n_code_links":2,"syntology":null},{"rank_in_archive_order":16,"model":"pix2pix","metrics":{"Class IOU":"0.18","LPIPS":"0","Per-class Accuracy":"25.0","Per-pixel Accuracy":"71.0"},"uses_additional_data":false,"paper_date":"2016-11-21","paper":"/paper/image-to-image-translation-with-conditional","paper_url":"http://arxiv.org/abs/1611.07004v3","paper_title":"Image-to-Image Translation with Conditional Adversarial Networks","code":"https://github.com/tensorflow/models/tree/master/research/gan","n_code_links":192,"syntology":{"n_ran":14,"n_unverified":108,"n_samples":122,"n_pointer_only_licence":1}},{"rank_in_archive_order":17,"model":"CycleGAN","metrics":{"Class IOU":" 0.11","Per-class Accuracy":"17%","Per-pixel Accuracy":"52%"},"uses_additional_data":false,"paper_date":"2017-03-30","paper":"/paper/unpaired-image-to-image-translation-using","paper_url":"https://arxiv.org/abs/1703.10593v7","paper_title":"Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks","code":"https://github.com/labmlai/annotated_deep_learning_paper_implementations","n_code_links":190,"syntology":{"n_ran":6,"n_unverified":25,"n_samples":31,"n_pointer_only_licence":6}},{"rank_in_archive_order":18,"model":"CoGAN","metrics":{"Class IOU":"0.06","Per-class Accuracy":"10%","Per-pixel Accuracy":"40%"},"uses_additional_data":false,"paper_date":"2016-06-24","paper":"/paper/coupled-generative-adversarial-networks","paper_url":"http://arxiv.org/abs/1606.07536v2","paper_title":"Coupled Generative Adversarial Networks","code":"https://github.com/eriklindernoren/PyTorch-GAN","n_code_links":4,"syntology":{"n_ran":0,"n_unverified":5,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"SimGAN","metrics":{"Class IOU":"0.04","Per-class Accuracy":"10%","Per-pixel Accuracy":"20%"},"uses_additional_data":false,"paper_date":"2016-12-22","paper":"/paper/learning-from-simulated-and-unsupervised","paper_url":"http://arxiv.org/abs/1612.07828v2","paper_title":"Learning from Simulated and Unsupervised Images through Adversarial Training","code":"https://github.com/carpedm20/simulated-unsupervised-tensorflow","n_code_links":9,"syntology":{"n_ran":2,"n_unverified":6,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"BiGAN","metrics":{"Class IOU":" 0.02","Per-class Accuracy":"6%","Per-pixel Accuracy":"19%"},"uses_additional_data":false,"paper_date":"2016-06-02","paper":"/paper/adversarially-learned-inference","paper_url":"http://arxiv.org/abs/1606.00704v3","paper_title":"Adversarially Learned Inference","code":"https://github.com/MaximeVandegar/Papers-in-100-Lines-of-Code/tree/main/Adversarially_Learned_Inference","n_code_links":9,"syntology":{"n_ran":2,"n_unverified":2,"n_samples":4,"n_pointer_only_licence":2}},{"rank_in_archive_order":21,"model":"INADE","metrics":{"LPIPS":"0.248"},"uses_additional_data":false,"paper_date":"2021-03-11","paper":"/paper/diverse-semantic-image-synthesis-via","paper_url":"https://arxiv.org/abs/2103.06878v1","paper_title":"Diverse Semantic Image Synthesis via Probability Distribution Modeling","code":"https://github.com/tzt101/INADE","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":2,"n_samples":5,"n_pointer_only_licence":5}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":9,"rows_with_any_sample_ran":8,"distinct_papers_with_graph_line":9,"distinct_papers_with_any_sample_ran":8,"samples_over_distinct_papers":{"n_ran":40,"n_unverified":156,"n_samples":196,"n_pointer_only_licence":26,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":40,"n_unverified":156,"n_samples":196,"n_pointer_only_licence":26,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}