【行为识别综述准备】
首先,以后可以工作可以考虑一下基于图卷及的行为识别今年很多,且在数据集上性能领先。
如下图所示:
应用场景:
- 【2019】Skeleton-based Action Recognition of People Handling Objects 【论文】
- recognizing object-related human actions
- 偏向于应用场景
- 通过构建skeletion-graph 利用了图卷积
CNN-Based:
- 【2019】SkeleMotion: A New Representation of Skeleton Joint Sequences Based on Motion Information for 3D Action Recognition【论文】 【代码】
- Skeleton -> Image
- 偏向于方法
- 3D 行为识别
- CNN方法
- NTU RGB+D 120 dataset
- skeleton image representation 重要!!!对骨架序列进行编码,进一步表示,高效表示,skeleton->Image,有点像师兄的思路!
2. 【2019】Three-Stream Convolutional Neural Network With Multi-Task and Ensemble Learning for 3D Action Recognition【论文】
- 数据集:NTU RGBD
- 3D 行为识别
- 思路:三个stage
- 偏向于方法,网络结构上的创新!
- 多任务学习
此外他还有一篇文章是考虑了频域信息进去的,结构差不多
如下,主要讲当前的方法都是异步学习语义信息等的,而且是在不同的层,这里提出了residual frequency domin的attention block!也是网络结构的上的改进!(多种模态信息的考虑,时-空-频率)
- Skeleton based action recognition using translation-scale invariant image mapping and multi-scale deep cnn 【论文】
- 对骨架信息转换成特殊形式的Image
- 多尺度的CNN
- 数据集: NTU RGB+D, UTD-MHAD, MSRC-12, and G3D
- 【2019】Making the Invisible Visible: Action Recognition Through Walls and Occlusions 【论文】
- 对黑暗中或者遮挡住的人体也能构建成骨架,然后识别动作,骨架信息作为中间信息!
- CNN
- 加入了Attention Module用来获取时空信息
- 【2018】Co-occurrence Feature Learning from Skeleton Data for Action Recognition and Detection with Hierarchical Aggregation
【论文】
- End- to-end
- CNN
- 数据集:NTU RGB+D, SBU Kinect Interaction and PKU-MMD.
- 首先每个点的信息被单独的进行了编码,其次将其assembled into到同时包含时空 信息的高层语音的representation形式,
- 【2019ICME】Learning Shape-Motion Representations from Geometric Algebra Spatio-Temporal Model for Skeleton-Based Action Recognition 【论文】
- 也是对骨架信息的编码,更高效的表示,然后送到CNN
- 利用了集合代数Geometric Algebra
- multi-stream CNN model
- 数据集:NTU RGB+D and Northwestern-UCLA datasets
- 【2017】3D CNNs on Distance Matrices for Human Action Recognition 【论文】
- 3DCNN+DM(欧式距离矩阵)来获取的良好的空间几何结构信息
- 数据集:差不多还是有NTU
- 比之前相近的LSTM-based的方法提升了10个点
- 也可以算是一种对于骨架信息的表示吧
- 【2017】Skeleton-based Action Recognition with Convolutional Neural Networks 【论文】
- CNN+Motion+Trans,还是将骨架序列当做图片去处理
- action classification and detection
- 骨架序列直接送进CNN之后通过骨架transformer模块自动选择中药的骨架点
- 数据集NTU PKUMMD
- 【PR2017】Enhanced skeleton visualization for view invariant human action recognition-师兄的论文! 【论文】
- Synthesized CNN
- 也是对骨架点进行转换!之后CNN
- 【2017】A New Representation of Skeleton Sequences for 3D Action Recognition 【论文】
- CNN
- 也是一种对骨架序列重新进行加工的形式使其更加具有表现性!
- A skeleton sequence of any length is transformed into three clips each consisting of several gray images. The generated clips are then fed to a deep CNN model to extract CNN features which are used in a MTLN for action recognition.
- 【2018】A Fine-to-Coarse Convolutional Neural Network for 3D Human Action Recognition 【论文】
- 对骨架序列进行分割成小片段,用来学习其中的联系
- 时空信息通过 fine-to-coarse (F2C) CNN architecture 实现
- 数据集:NTU RGB+D and SBU Kinect Interaction dataset.
- 也是对骨架信息动手脚
- 【2017】Interpretable 3D Human Action Analysis with Temporal Convolutional Networks
- CNN
- a way to explicitly learn readily interpretable spatio-temporal representations for 3D human action recognition.
- 【2017】SkeletonNet: Mining Deep Part Features for 3D Action Recognition 【论文】
- 也是一种表现实行,相比于骨架信息本身,处理后的包含了旋转,评议,和尺度等因素
- 由两部分构成,第一个用于提取特征,另一个用来将特征转化为更加具有区分性且紧凑的表现形式
- Two-Stream 3D Convolutional Neural Network for Skeleton-Based Action Recognition-涂涂师姐的论文
- 双流3DCNN
- 也是一种表现形式,对骨架点编码到利群上的点!
- 【2019】Skeleton Image Representation for 3D Action Recognition based on Tree Structure and Reference Joints 【论文】 【代码】
- a novel skeleton image representation to be used as input to CNNs.
- 数据集:NUTRGBD
- 也是对骨架信息进行处理然后送进CNN
- [2014] Skeletal Quads: Human Action Recognition Using Joint Quadruples 【论文】
- 对骨架信息进行编码!
- 没准对之后的工作又用!
- Ensemble One-dimensional Convolution Neural Networks for Skeleton-based Action Recognition
网络结构的改变 - Hard Sample Mining and Learning for Skeleton-Based Human Action Recognition and Identification 【论文】
- 行为识别+人的识别identification(应用!)
- 特点:轻量级,快!
混合方法;
- 【2019】Semantics-Guided Neural Networks for Efficient Skeleton-Based Human Action Recognition [【论文】] (https://arxiv.org/pdf/1904.01189v1.pdf)
- e NTU, SYSU, and NUCLA datasets.
- GCN+ CNN
- 也是获取多种模态信息,如图
- 【ECCV2018】Skeleton-Based Action Recognition with Spatial Reasoning and Temporal Stack Learning 【论文】 【代码】
- 图卷积+LSTM相关,前者获取高层空间结构信息,后者获取时域动态信息
- propose a novel model with spatial reasoning and temporal stack learning (SR-TSL) for skeleton based action recognition
RNN-based
1.【2018】 Memory Attention Networks for Skeleton-based Action Recognition 【论文】 【代码】
- 特点:In this work, we propose a temporal-then-spatial recalibration scheme to alleviate such complex variations, resulting in an end-to-end Memory Attention Networks (MANs) which consist of a Temporal Attention Recalibration Module (TARM) and a Spatio-Temporal Convolution Module (STCM).
- 数据集:NTU RGB+D, HDM05, SYSU-3D and UT-Kinec