论文阅读《Long-term Temporal Convolutions for Action Recognition》
来源:互联网 发布:琉璃神社新的域名 编辑:程序博客网 时间:2024/05/17 23:48
论文阅读《Long-term Temporal Convolutions for Action Recognition》
标签(空格分隔): ActionRecognition ReadingNote
project page :http://www.di.ens.fr/willow/research/ltc
gayhub: https://github.com/gulvarol/ltc
现有方法一般采用CNN来获取动作表达,但是一般来说从较少的视频帧中,很难完整的对一个动作建模。
本文工作:
- LTC(long-term temporal convolutions)
- 研究了不同的low-level表达(比如视频原始像素、光流向量)的影响
- 强调高质量光流估计的重要性。
实验结果:UCF101 (92.7%) and HMDB51 (67.2%)。
Intro
Related Work
LONG-TERM TEMPORAL CONVOLUTIONS
网络结构
The network has 5 space-time convolutional layers with 64, 128, 256, 256 and 256 filter response maps, followed by 3 fully connected layers of sizes 2048, 2048 and number of classes.
卷积核3x3x3,每个卷积层都采用ReLU激活函数,以及max-pooling(除了第一层是2x2x1,其他都是2x2x2)。三个维度都padding一个像素,卷积stride 1 for conv,2 for pooling。前两个fc层采用dropout。fc层结束后都接上ReLU层,最后采用softmax层输出类别分数。
网络输入
从C3D出发,首先对比16帧和60帧的输入。
系统分析输入信号的时空分辨率增长的影响(两个方面motion/appearance)。
16帧:112x112x16 crop,171x128像素
60帧:58x58x60 crop,89x67像素
60帧的网络,5个卷积层分别对应60、30、15、7、3帧输入。
16:16、8、4、2、1.
The space-time resolution for the outputs of the fifth convolutional layers is
For a systematic study of networks with different input resolutions we also evaluate the effect of increased temporal resolution t 2 f20; 40; 60; 80; 100g and varying spatial resolution of
除了采用RGB输入,还采用了x和y方向的光流输入。
三种光流:MPEG flow、Farneback、Brox.
learning
一些训练细节。
dataset:HMDB51(3.7k)/UCF101(9.5k)
SGD,mini-batch,对数似然。
16f network:30 clips/batch
60f network:15clips/batch
100f network:10clips/batch
初始学习率:
from scratch:
UCF101,减小两次,每次除以10。( 16f:迭代80k减小一次,再迭代45k减小一次,再20k多迭代就差不多了。)
HMDB51收敛快一些,60k时除以10就行了。在迭代10k多就差不多了。
对于60f network这些参数要double,100f要triple。
dropout:0.9
The momentum is set to 0.9
weight decay is initialized with
data augmentation:
- random clipping
- multiscale cropping
论文挺简单的……空了再写笔记吧。
- 论文阅读《Long-term Temporal Convolutions for Action Recognition》
- 【论文阅读笔记】CVPR2015-Long-term Recurrent Convolutional Networks for Visual Recognition and Description
- [论文阅读]Rank Pooling for Action Recognition
- [深度学习论文笔记][Video Classification] Long-term Recurrent Convolutional Networks for Visual Recognition a
- 论文笔记:Evaluation of local spatio-temporal features for action recognition
- 论文笔记-Temporal segment network:towards good practices for deep action recognition
- 论文笔记一Temporal Segment Networks: Towards Good Practices for Deep Action Recognition
- 论文阅读《Spatiotemporal Multiplier Networks for Video Action Recognition》
- 论文阅读《Dynamic Image Networks for Action Recognition》
- long term recurrent convolutional networks for visual recognition and description
- Long-term Recurrent Convolutional Networks for Visual Recognition and Description
- Long-term Recurrent Convolutional Networks for Visual Recognition and Description
- Learning hierarchical spatio-temporal features for action recognition with ISA
- 双流网络行为识别-Spatiotemporal Residual Networks for Video Action Recognition-论文阅读
- [论文阅读笔记]Two-Stream Convolutional Networks for Action Recognition in Videos
- [论文阅读笔记]Two-Stream Convolutional Networks for Action Recognition in Videos
- 【论文阅读】 输入法相关论文二 LONG SHORT TERM MEMORY NEURAL NETWORK
- Contextual Action Recognition with R*CNN-论文阅读
- CDN技术详解
- 25、Selenium + Python 实现 UI 自动化测试-unknown error: Element is not clickable at point (x,y)
- 源码解析-post
- 面试时关于“工作可预见困难有哪些”应聘者如何巧妙回答
- android读写excel文件
- 论文阅读《Long-term Temporal Convolutions for Action Recognition》
- python读取外部数据之读取csv格式
- 自动账户确定之AUM
- 图的数据结构
- reorder-list Java code
- continue;go to;;break;return
- MyBatis 接收数据库中没有的字段(记一次工作总结)
- Centos上ShadowSocks搭建Socket5服务器
- POJ 3057:Evacuation (搜索+二分图匹配)