{"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/two-stream-convolutional-networks-for-dynamic","title":"Two-Stream Convolutional Networks for Dynamic Texture Synthesis","arxiv_id":"1706.06982","date":"2017-06-21","proceeding":"CVPR 2018 6","authors":["Matthew Tesfaldet","Marcus A. Brubaker","Konstantinos G. Derpanis"],"abstract":"We introduce a two-stream model for dynamic texture synthesis. Our model is\nbased on pre-trained convolutional networks (ConvNets) that target two\nindependent tasks: (i) object recognition, and (ii) optical flow prediction.\nGiven an input dynamic texture, statistics of filter responses from the object\nrecognition ConvNet encapsulate the per-frame appearance of the input texture,\nwhile statistics of filter responses from the optical flow ConvNet model its\ndynamics. To generate a novel texture, a randomly initialized input sequence is\noptimized to match the feature statistics from each stream of an example\ntexture. Inspired by recent work on image style transfer and enabled by the\ntwo-stream model, we also apply the synthesis approach to combine the texture\nappearance from one texture with the dynamics of another to generate entirely\nnovel dynamic textures. We show that our approach generates novel, high quality\nsamples that match both the framewise appearance and temporal evolution of\ninput texture. Finally, we quantitatively evaluate our texture synthesis\napproach with a thorough user study.","url_abs":"http://arxiv.org/abs/1706.06982v4","url_pdf":"http://arxiv.org/pdf/1706.06982v4.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":"two-stream-convolutional-networks-for-dynamic","repo_url":"https://github.com/ryersonvisionlab/two-stream-dyntex-synth","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"object-recognition","task_name":"Object Recognition"},{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"},{"task_slug":"style-transfer","task_name":"Style Transfer"},{"task_slug":"texture-synthesis","task_name":"Texture Synthesis"},{"task_slug":"two","task_name":"Vocal Bursts Valence Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.06982","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}