Joint-Aware Action Recognition for Ambient Assisted Living

Oikonomou, Katerina Maria; Kansizoglou, Ioannis; Manaveli, Pelagia; Grekidis, Athanasios; Menychtas, Dimitrios; Aggelousis, Nikolaos; Sirakoulis, Georgios Ch.; Γαστεράτος, Αντώνιος

doi:10.1109/ist55454.2022.9827760

Cited by 10 publications

(11 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although this checksum method was introduced for a convolution between one IFM and a single kernel, its properties can be used to calculate the checksum of a whole CNN layer. In the general case, the explicit checksum for a whole CNN layer can be calculated using (2), where a (k,l) i,j,n is a feature located at row i and column j of the n IFM and that during the convolution it gets multiplied with a weight that is located in row k and column l of a kernel. The sum of all a (k,l) i,j,n is then multiplied with the sum of all weights with indexes k, l, n across all the M filters.…”

Section: E Integrated Checksum Checkermentioning

confidence: 99%

“…In the third scenario, we opted for a technique tailored for recognizing human activities based on joint awareness [2]. In this particular instance, a 2-dimensional CNN architecture is utilized to categorize five typical daily activities in indoor settings using 3-D skeletal data.…”

Section: ) Human-activity Recognitionmentioning

confidence: 99%

“…In general, the need for custom CNN accelerators, that can easily adapt to the requirements of each application, has been increased as more applications have emerged, ranging from low power wearable devices to high performance data centers. Such a necessity is especially highlighted in assisted living systems developed for domestic environments, where the provided deep learning technologies, i.e., human pose estimation [1], action recognition [2] and fall detection [3], have to be endowed with low-energy consumption capacities. Therefore, to cover the wide variety of CNN applications and their constraints, the design of custom hardware accelerators has to be easily accessible and configurable.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A High-Level Synthesis Library for Synthesizing Efficient and Functional-Safe CNN Dataflow Accelerators

Filippas,

Peltekis,

Titopoulos

et al. 2024

IEEE Access

Self Cite

View full text Add to dashboard Cite

Convolution neural networks (CNNs) are widely applied in many machine learning applications. Hardware acceleration for CNNs is crucial, given their high computational intensity and the demand for enhanced energy efficiency and reduced latency in application response. This work leverages the simplicity of modelling CNN structure in Python with the flexibility of High-Level synthesis to automate the creation of CNN dataflow hardware accelerators. The methodology emphasizes ease of design, enabling users to effortlessly generate hardware accelerators directly from the corresponding Tensorflow or PyTorch CNN models. The proposed generator incorporates multiple optimization knobs, facilitating the automatic exploration of diverse architectural alternatives to enhance performance and improve energy efficiency. Besides well-known optimization knobs, this work integrates for the first time functional safety features and optimized buffering architectures. It also introduces various design options related to data representation and arithmetic. Experimental results highlight (a) the effectiveness of the proposed approach in automatically synthesizing optimized CNN dataflow accelerators and (b) the improved characteristics offered by the introduced generator compared to state-of-the-art approaches.

show abstract

Section: E Integrated Checksum Checkermentioning

confidence: 99%

Section: ) Human-activity Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A High-Level Synthesis Library for Synthesizing Efficient and Functional-Safe CNN Dataflow Accelerators

Filippas,

Peltekis,

Titopoulos

et al. 2024

IEEE Access

Self Cite

View full text Add to dashboard Cite

show abstract

“…Some methods re-weight the joint data to produce better representations. They might be motivated for a variety of reasons, such as to solve the sample imbalance problem [21] or discovering the discriminative joints [22]. The network is able to better capture temporal and spatial information [23] by using the attention mechanism, which also helps the network focus attention on the important areas [24].…”

Section: Part-based Skeleton Action Recognitionmentioning

confidence: 99%

Part-Wise Adaptive Topology Graph Convolutional Network for Skeleton-Based Action Recognition

et al. 2023

View full text Add to dashboard Cite

Human action recognition is a computer vision challenge that involves identifying and classifying human movements and activities. The behavior of humans comprises movements of multiple body parts, and Graph Convolutional Networks (GCNs) have emerged as a promising approach for this task. However, most contemporary GCN methods perform graph convolution on the entire skeleton graph without considering that the human body consists of distinct body parts. To address these shortcomings, we propose a novel method that optimizes the representation of the skeleton graph by designing temporal and spatial convolutional blocks while introducing the Part-wise Adaptive Topology Graph Convolution (PAT-GC) technique. PAT-GC adaptively learns the segmentation of different body parts and dynamically integrates the spatial relevance between them. Furthermore, we utilize hierarchical modeling to divide the skeleton graph, capturing a more comprehensive representation of the human body. We evaluate our approach on three publicly available large datasets: NTU RGB + D 60, NTU RGB + D 120, and Kinetics Skeleton 400. Our experimental results demonstrate that our approach achieves state-of-the-art performance, thus validating the efficiency of our proposed technique for human action recognition.

show abstract

“…All these systems, along with many other commercial solutions, rely on camera-based surveillance and often incorporate image processing techniques and artificial intelligence to improve overall accuracy. Some also consider the energy efficiency of video surveillance systems, as discussed in [16,17].…”

Section: Introductionmentioning

confidence: 99%

Exploring Older Adults’ Willingness to Install Home Surveil-Lance Systems in Taiwan: Factors and Privacy Concerns

Wang

Lin

2023

Healthcare

View full text Add to dashboard Cite

Taiwan has a rapidly increasing aging population with a considerably high life expectancy rate, which poses challenges for healthcare and medical systems. This study examines three key factors: safety concerns, family expectations, and privacy concerns, and their influence on surveillance system installation decisions. A cross-sectional study was conducted involving a group of physically active older adults in Taiwan, using a questionnaire to collect data on the reasons for in-stalling a surveillance system and preferences for three image privacy protection techniques: blurring the face and transformation to a 2D or 3D character. The study concluded that while safety concerns and family expectations facilitate the adoption of surveillance systems, privacy concerns serve as a significant barrier. Furthermore, older adults showed a clear preference for avatar-based privacy protection methods over simpler techniques, such as blurring. The outcomes of this research will be instrumental in shaping the development of privacy-conscious home surveillance technologies, adeptly balancing safety and privacy. This understanding can pave the way for technology design that skillfully balances privacy concerns with remote monitoring quality, thereby enhancing the well-being and safety of this demographic. These results could possibly be extended to other demographics as well.

show abstract

Joint-Aware Action Recognition for Ambient Assisted Living

Cited by 10 publications

References 13 publications

A High-Level Synthesis Library for Synthesizing Efficient and Functional-Safe CNN Dataflow Accelerators

A High-Level Synthesis Library for Synthesizing Efficient and Functional-Safe CNN Dataflow Accelerators

Part-Wise Adaptive Topology Graph Convolutional Network for Skeleton-Based Action Recognition

Exploring Older Adults’ Willingness to Install Home Surveil-Lance Systems in Taiwan: Factors and Privacy Concerns

Contact Info

Product

Resources

About