{"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/ev-flownet-self-supervised-optical-flow","title":"EV-FlowNet: Self-Supervised Optical Flow Estimation for Event-based Cameras","arxiv_id":"1802.06898","date":"2018-02-19","proceeding":null,"authors":["Alex Zihao Zhu","Liangzhe Yuan","Kenneth Chaney","Kostas Daniilidis"],"abstract":"Event-based cameras have shown great promise in a variety of situations where\nframe based cameras suffer, such as high speed motions and high dynamic range\nscenes. However, developing algorithms for event measurements requires a new\nclass of hand crafted algorithms. Deep learning has shown great success in\nproviding model free solutions to many problems in the vision community, but\nexisting networks have been developed with frame based images in mind, and\nthere does not exist the wealth of labeled data for events as there does for\nimages for supervised training. To these points, we present EV-FlowNet, a novel\nself-supervised deep learning pipeline for optical flow estimation for event\nbased cameras. In particular, we introduce an image based representation of a\ngiven event stream, which is fed into a self-supervised neural network as the\nsole input. The corresponding grayscale images captured from the same camera at\nthe same time as the events are then used as a supervisory signal to provide a\nloss function at training time, given the estimated flow from the network. We\nshow that the resulting network is able to accurately predict optical flow from\nevents only in a variety of different scenes, with performance competitive to\nimage based networks. This method not only allows for accurate estimation of\ndense optical flow, but also provides a framework for the transfer of other\nself-supervised methods to the event-based domain.","url_abs":"http://arxiv.org/abs/1802.06898v4","url_pdf":"http://arxiv.org/pdf/1802.06898v4.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":"ev-flownet-self-supervised-optical-flow","repo_url":"https://github.com/daniilidis-group/EV-FlowNet","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"ev-flownet-self-supervised-optical-flow","repo_url":"https://github.com/Cyril-Sterling/EVFlowNet-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"optical-flow-estimation","task_name":"Optical Flow Estimation"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1802.06898","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}