188宝金博页面版

  • 图案背景
  • 纯色背景
视图
标记
批注
批注本地保存成功，开通会员云端永久保存 去开通
adsuhviusa

上传于：2018-08-28

粉丝量：104

该文档贡献者很忙，什么也没留下。


  • 相关
  • 目录
  • 笔记
  • 书签

188宝金博页面版:更多相关文档

  • Future's future

    星级： 3 页

  • future 将来 未来 in future 今后 in the future 在将来

    星级： 3 页

  • Futures' futures

    星级： 1 页

  • in future与in the future

    星级： 1 页

  • The future of futures

    星级： 12 页

  • Future and Near Future

    星级： 5 页

  • In the future, in the future, and in the future

    星级： 12 页

  • The future is in the future

    星级： 12 页

  • future

    星级： 10 页

  • ~future

    星级： 2 页

  • VI – FUTURES; FUTURS

    星级： 2 页

暂无目录

点击鼠标右键菜单,创建目录

    暂无笔记

    选择文本，点击鼠标右键菜单，添加笔记

    暂无书签

    在左侧文档中，点击鼠标右键，添加书签

    188宝金博页面版: Future Semantic Segmentation with Convolutional LSTM

    下载积分： 500

    内容提示： NABAVI, ROCHAN, WANG: FUTURE SEMANTIC SEGMENTATION 1Future Semantic Segmentation withConvolutional LSTMSeyed shahabeddin Nabavinabaviss@cs.umanitoba.caMrigank Rochanmrochan@cs.umanitoba.caYang Wangywang@cs.umanitoba.caDepartment of Computer ScienceUniversity of ManitobaWinnipeg, MB, CanadaAbstractWe consider the problem of predicting semantic segmentation of future frames ina video. Given several observed frames in a video, our goal is to predict the semanticsegmentation map of future frames that are not y...

    文档格式：PDF | 页数：12 | 浏览次数：29 | 上传日期：2018-08-28 18:48:22 | 文档星级：
    NABAVI, ROCHAN, WANG: FUTURE SEMANTIC SEGMENTATION 1Future Semantic Segmentation withConvolutional LSTMSeyed shahabeddin Nabavinabaviss@cs.umanitoba.caMrigank Rochanmrochan@cs.umanitoba.caYang Wangywang@cs.umanitoba.caDepartment of Computer ScienceUniversity of ManitobaWinnipeg, MB, CanadaAbstractWe consider the problem of predicting semantic segmentation of future frames ina video. Given several observed frames in a video, our goal is to predict the semanticsegmentation map of future frames that are not yet observed. A reliable solution to thisproblem is useful in many applications that require real-time decision making, such asautonomous driving. We propose a novel model that uses convolutional LSTM (Con-vLSTM) to encode the spatiotemporal information of observed frames for future pre-diction. We also extend our model to use bidirectional ConvLSTM to capture temporalinformation in both directions. Our proposed approach outperforms other state-of-the-artmethods on the benchmark dataset.1 IntroductionWe consider the problem of future semantic segmentation in videos. Given several framesin a video, our goal is to predict the semantic segmentation of unobserved frames in thefuture. See Fig. 1 for an illustration of the problem. The ability to predict and anticipatethe future plays a vital role in intelligent system decision-making [3, 21]. An example isthe autonomous driving scenario. If an autonomous vehicle can correctly anticipate thebehaviors of other vehicles [5] or predict the next event that will happen in accordance withthe current situation (e.g. collision prediction [1]), it can take appropriate actions to preventdamages.Computer vision has made signif icant progress in the past few years. However, moststandard computer vision tasks (e.g. object detection, semantic segmentation) focus on pre-dicting labels on images that have been observed. Predicting and anticipating the future isstill challenging for current computer vision systems. Part of the challenge is due to theinherent uncertainty of this problem. Given one or more observed frames in a video, thereare many possible events that can happen in the future.There has been a line of research on predicting raw RGB pixel values of future framesin a video sequence [10, 14, 17, 20]. While predicting raw RGB values of future frames isuseful, it may not be completely necessary for downstream tasks. Another line of researchc ? 2018. The copyright of this document resides with its authors.It may be distributed unchanged freely in print or electronic forms.arXiv:1807.07946v1 [cs.CV] 20 Jul 2018

    阅读了该文档的用户还阅读了这些文档

    188宝金博页面版:关于我们

    188宝金博页面版:帮助中心

    188宝金博页面版:关注我们

    • 新浪微博

    关注188宝金博页面版公众号

    188宝金博页面版
    阅读
    APP
    阅读
    返回
    顶部
    188宝金博页面版官网登录在线平台入口（2026已更新）—江苏协昌电子科技股份有限公司