TEINet: Towards an Efficient Architecture for Video Recognition(AAAI2020)
AAAI2020 南京大学+腾讯优图实验室 1 摘要 在视频动作识别的架构设计中，效率是一个重要的问题。3D CNNs在视频动作识别方面取得了显著的进展。然而，与二维卷积相比，三维卷积往往引入大量的参数，导致计算量大。为了解决这个问题，我们提出了一个有效的时序模块，称为Temporal Enhancement-and-Interaction（TEI模块），它可以插入到现有的2D CNNs中。TEI
相关文章
- 1. 论文浏览(22) TEINet: Towards an Efficient Architecture for Video Recognition
- 2. Video Architecture Search
- 3. ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
- 4. 6 Multi-Cloud Architecture Designs for an Effective Cloud
- 5. Path-Level Network Transformation for Efficient Architecture Search
- 6. ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
- 7. Paper Reading：ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
- 8. Flash: An Efficient and Portable Web Server
- 9. Spatially Adaptive Residual Networks for Efficient Image and Video Deblurring
- 10. ForkBase: An Efficient Storage Engine for Blockchain and Forkable Applications
- 更多相关文章...
- • Swift for 循环 - Swift 教程
- • Scala for循环 - Scala教程
- • RxJava操作符（一）Creating Observables
- • PHP开发工具
相关标签/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
欢迎关注本站公众号,获取更多信息
相关文章
- 1. 论文浏览(22) TEINet: Towards an Efficient Architecture for Video Recognition
- 2. Video Architecture Search
- 3. ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices
- 4. 6 Multi-Cloud Architecture Designs for an Effective Cloud
- 5. Path-Level Network Transformation for Efficient Architecture Search
- 6. ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
- 7. Paper Reading：ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design
- 8. Flash: An Efficient and Portable Web Server
- 9. Spatially Adaptive Residual Networks for Efficient Image and Video Deblurring
- 10. ForkBase: An Efficient Storage Engine for Blockchain and Forkable Applications