{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/crossnet-an-end-to-end-reference-based-super","title":"CrossNet: An End-to-end Reference-based Super Resolution Network using Cross-scale Warping","arxiv_id":"1807.10547","date":"2018-07-27","proceeding":"ECCV 2018 9","authors":["Haitian Zheng","Mengqi Ji","Haoqian Wang","Yebin Liu","Lu Fang"],"abstract":"The Reference-based Super-resolution (RefSR) super-resolves a low-resolution\n(LR) image given an external high-resolution (HR) reference image, where the\nreference image and LR image share similar viewpoint but with significant\nresolution gap x8. Existing RefSR methods work in a cascaded way such as patch\nmatching followed by synthesis pipeline with two independently defined\nobjective functions, leading to the inter-patch misalignment, grid effect and\ninefficient optimization. To resolve these issues, we present CrossNet, an\nend-to-end and fully-convolutional deep neural network using cross-scale\nwarping. Our network contains image encoders, cross-scale warping layers, and\nfusion decoder: the encoder serves to extract multi-scale features from both\nthe LR and the reference images; the cross-scale warping layers spatially\naligns the reference feature map with the LR feature map; the decoder finally\naggregates feature maps from both domains to synthesize the HR output. Using\ncross-scale warping, our network is able to perform spatial alignment at\npixel-level in an end-to-end fashion, which improves the existing schemes both\nin precision (around 2dB-4dB) and efficiency (more than 100 times faster).","url_abs":"http://arxiv.org/abs/1807.10547v1","url_pdf":"http://arxiv.org/pdf/1807.10547v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"crossnet-an-end-to-end-reference-based-super","repo_url":"https://github.com/htzheng/ECCV2018_CrossNet_RefSR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"patch-matching","task_name":"Patch Matching"},{"task_slug":"reference-based-super-resolution","task_name":"Reference-based Super-Resolution"},{"task_slug":"super-resolution","task_name":"Super-Resolution"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1807.10547","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}