Human Activity Recognition (HAR) Using Deep Learning: Review, Methodologies, Progress and Future Research Directions

Kumar, Pranjal; Chauhan, Siddhartha; Awasthi, Lalit Kumar

doi:10.1007/s11831-023-09986-x

Cited by 26 publications

(2 citation statements)

References 307 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, with the rise of attention mechanisms, integrating them into hybrid models has become a significant trend. Attention mechanisms allow models to focus more on the important parts of the data, thereby enhancing HAR accuracy and efficiency [31]. Mim et al proposed an efficient recognition model [32] that utilizes GRUs with an attention mechanism to extract temporal features and inception with a convolutional block attention module (CBAM) for spatial feature extraction, and it includes residual connections to address gradient vanishing issues.…”

Section: Introductionmentioning

confidence: 99%

Wearable Sensor-Based Residual Multifeature Fusion Shrinkage Networks for Human Activity Recognition

Zeng,

Guo,

Tan

et al. 2024

Sensors

View full text Add to dashboard Cite

Human activity recognition (HAR) based on wearable sensors has emerged as a low-cost key-enabling technology for applications such as human–computer interaction and healthcare. In wearable sensor-based HAR, deep learning is desired for extracting human active features. Due to the spatiotemporal dynamic of human activity, a special deep learning network for recognizing the temporal continuous activities of humans is required to improve the recognition accuracy for supporting advanced HAR applications. To this end, a residual multifeature fusion shrinkage network (RMFSN) is proposed. The RMFSN is an improved residual network which consists of a multi-branch framework, a channel attention shrinkage block (CASB), and a classifier network. The special multi-branch framework utilizes a 1D-CNN, a lightweight temporal attention mechanism, and a multi-scale feature extraction method to capture diverse activity features via multiple branches. The CASB is proposed to automatically select key features from the diverse features for each activity, and the classifier network outputs the final recognition results. Experimental results have shown that the accuracy of the proposed RMFSN for the public datasets UCI-HAR, WISDM, and OPPORTUNITY are 98.13%, 98.35%, and 93.89%, respectively. In comparison with existing advanced methods, the proposed RMFSN could achieve higher accuracy while requiring fewer model parameters.

show abstract

Section: Introductionmentioning

confidence: 99%

Wearable Sensor-Based Residual Multifeature Fusion Shrinkage Networks for Human Activity Recognition

Zeng,

Guo,

Tan

et al. 2024

Sensors

View full text Add to dashboard Cite

show abstract

“…Combined with the development and innovation of science and technology, a new era is ushering in the field of assessment and correction of dancers' dance posture in the field of dance art (Kumar et al, 2024). Through the use of advanced technological means, dancers' performance level and technical level can be improved to promote the continuous development and growth of dance art.…”

mentioning

confidence: 99%

The Role of the Combination of 3D Simulation Sequence Diagram and Video Motion Recognition Technology in Evaluating and Correcting Dancers' Dance Moves

Wei,

Chau

2024

International Journal of Mobile Computing and Multimedia Communications

View full text Add to dashboard Cite

This paper combines the dance 3D space simulation sequence diagram with video motion recognition technology, filters, denoises, grays and background removal the collected dance video images, analyzes the motion characteristics of people in the sequence diagram, uses support vector machine to learn and train 3D space models, classifies and recognizes people's dance movements, and extracts 3D-SIFT and optical flow characteristics of various areas of human body. Form a three-dimensional space simulation sequence diagram, reduce and normalize the extracted features, get the feature vectors of various characters, and input them into the classifier to realize the recognition of dance movements. The results show that the combination of 3D-SIFT and optical flow can realize the dynamic change of human static information, the illumination invariance of SIFT features can make up for the illumination sensitivity of optical flow features, and the optical flow features can solve the instability problem of determining the key points of SIFT features.

show abstract

Image-Based Human Action Recognition with Transfer Learning Using Grad-CAM for Visualization

Alam,

Acquaah,

Roy

2024

IFIP Advances in Information and Communication Technology

View full text Add to dashboard Cite

Human Activity Recognition (HAR) Using Deep Learning: Review, Methodologies, Progress and Future Research Directions

Cited by 26 publications

References 307 publications

Wearable Sensor-Based Residual Multifeature Fusion Shrinkage Networks for Human Activity Recognition

Wearable Sensor-Based Residual Multifeature Fusion Shrinkage Networks for Human Activity Recognition

The Role of the Combination of 3D Simulation Sequence Diagram and Video Motion Recognition Technology in Evaluating and Correcting Dancers' Dance Moves

Image-Based Human Action Recognition with Transfer Learning Using Grad-CAM for Visualization

Contact Info

Product

Resources

About