201904Online Human Action Recognition Based on Incremental Learning of Weighted Covariance Descriptors

论文标题:Online Human Action Recognition Based on Incremental Learning of Weighted Covariance Descriptors

来源/作者机构情况:

卧龙岗大学,第一次听说这个学校。竟然是在澳大利亚的一个学校。好吧,华人果然全球了

解决问题/主要思想贡献:

使用一个加权协方差因子,来积累前几帧的信息,使用增强学习来实现online learning,可以不用使用分好段的视频来预测动作

成果/优点:

缺点:

反思改进/灵感:

#############################################################

论文主要内容与关键点:

1.Introduction

前人研究的主要分类方法,缺点是没有办法实时检测

视频动作的特征表现,主要依靠这两种

2.Related Work

主要介绍了一下,上面两种分类方法,主要的几个研究方法,讲了一下这些的缺点。

特别强调了一个苏联人的一个方法,并讲解自己的文字解决了他的两个问题:没有权重的对待不同帧

3.The proposed method


权重方差因子:

时间权重的变化:

帧权重的变化:

这里,能量是结点的动量和势能。

增量学习:

后面还给予了证明。

4.Experimental Results

一些度量性能。还有结果展示

后面使用了一些去噪和归一化

动作特征向量的选取:

使用KNN和SVM进行特征分类

展示了在三个数据集上面的结果:

最后一种数据集,是卧龙岗大学自己创造的,适合增量学习。

5The depth and skeleton data will be made available after the paper being accepted for publication at:

http://www.uow.edu.au/˜wanqing/#UOWActionDatasets

https://www.uow.edu.au/~wanqing/#UOWActionDatasets

最后讲解了使用的硬件情况:

以及对比,KNN和SVM的情况:

5.Conclusion

6.附录,

一些公式的证明和推导

代码实现:

原文地址:https://www.cnblogs.com/captain-dl/p/10759693.html

时间: 2024-11-05 12:23:02

201904Online Human Action Recognition Based on Incremental Learning of Weighted Covariance Descriptors的相关文章

Spatio-temporal feature extraction and representation for RGB-D human action recognition

propose a novel and effective framework to largely improve the performance of human action recognition using both the RGB videos and depth maps. The key contribution is the proposition of the sparse coding-based temporal pyramid matching approach (Sc

Human Action Recognition Using Kinect

Abstract: In this paper, a general approach for human action recognition is applied for classifying human movements into action classes . The propose method uses Kinect for capturing depth  stream. The system performs preprocessing on depth informati

【论文笔记】Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition

Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition 2018-01-28  15:45:13  研究背景和动机: 行人动作识别(Human Action Recognition)主要从多个模态的角度来进行研究,即:appearance,depth,optical-flow,以及 body skeletons.这其中,动态的人类骨骼点 通常是最具有信息量的,且能够和其他模态进行互补.

COMP255 Human activity recognition

Assignment: SDLC for human activityrecognition projectCOMP255 Human activity recognition @MQ S2, 2019 (updated on 11thAug 2019)This project aims to develop a human activity recognition IoT application toevaluate students’ knowledge in SDLC. This is a

Review of Image Super-resolution Reconstruction Based on Deep Learning

Abstract With the deep learning method being applied to image super-resolution (SR), SR methods based on deep learning have achieved better reconstruction results than traditional SR methods. This paper briefly summarizes SR methods based on deep lea

Progress and Prospect of target detection technology based on deep learning

Viola-Jones face detector One of the more successful examples of object detection in the whole computer field is the Viola-Jones face detector that appeared around 2000,which makes it a more mature technique compared to the object detection.The basic

[论文理解] Attentional Pooling for Action Recognition

Attentional Pooling for Action Recognition 简介 这是一篇NIPS的文章,文章亮点是对池化进行矩阵表示,使用二阶池的矩阵表示,并将权重矩阵进行低秩分解,从而使分解后的结果能够自底向上和自顶向下的解释,并巧用attention机制来解释,我感觉学到了很多东西,特别是张量分解等矩阵论的知识点. 基础概念 低秩分解 目的:去除冗余并减少模型的权值参数 方法:使用两个K*1的卷积核代替掉一个K*K的卷积核 原理:权值向量主要分布在一些低秩子空间,使用少量的基就可

Redis Installation、Configuration、Program Based On Redis Learning

目录 1. Redis 简介 2. Redis安装配置 3. 编程使用Redis 1. Redis 简介 0x1: Redis是什么 Redis是一款Nosql类型的基于key-value的高速缓存系统, 从架构上看,redis有3种特性 1. key value store 是一个以key-value形式存储的数据库,定位直指MySQL,用来作为唯一的存储系统 2. memory cache 是一个把数据存储在内存中的高速缓存,用来在应用和数据库间提供缓冲,替代memcachd 3. data

From N to N+1: Multiclass Transfer Incremental Learning 代码分析(1)

首先推荐大家看文章:From N to N+1: Multiclass Transfer Incremental Learning 核心细想就是Transfer Incremental Learning 原理图如下:Transfer learning的任务就是检测出小狗,并且系统已经学习到了几种动物(小猫.马).从有n个类别学习第n+1个类别. 代码组成部分: Contents -------- data/ -- demo data tmp/ -- temporary files (e.g. s