DeepSeg: Deep-Learning-Based Activity Segmentation Framework for Activity Recognition Using WiFi

Xiao, Chunjing; Yue, Lei; Ma, Yongsen; Zhang, Fan; Qin, Zhiguang

doi:10.1109/jiot.2020.3033173

Cited by 55 publications

(13 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although the designed RF-Eletter could achieve individual letter recognition in different domains, word recognition proved challenging following the highly complex and fine-grained gesture recognition of words (an intriguing complexity to be examined in future research). As such, data segmentation and subsequent techniques [62] could be considered to split the words into different parts and splice the content for word recognition. Different signal waveforms could also be generated for different words through relevant word recognition analysis.…”

Section: Discussionmentioning

confidence: 99%

RF-Eletter: A Cross-Domain English Letter Recognition System Based on RFID

Yang¹,

Liu

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Section: Discussionmentioning

confidence: 99%

RF-Eletter: A Cross-Domain English Letter Recognition System Based on RFID

Yang¹,

Liu

et al. 2021

IEEE Access

View full text Add to dashboard Cite

“…At the same time, due to the influence of environment and subjective consciousness, the setting of ordinary threshold may lead to a significant decline in the performance of mixed activities. In order to deal with these problems, inspired by the method in [ 36 ], we first segment the motion curve of continuous persons through its data volatility, and then use the sliding window segmentation method based on variance.…”

Section: Methodsmentioning

confidence: 99%

Capturing Features and Performing Human Detection from Human Gaits Using RFID

Zhang

Liu

Yang

et al. 2022

Sensors

View full text Add to dashboard Cite

Recently, radio frequency identification (RFID) sensing has attracted much attention due to its contact-free nature, low cost, light weight and other advantages. RFID-based person detection has also become a hot research topic, but there are still some problems in the existing research. First, most of the current studies cannot identify numerous people at a time well. Second, in order to detect more accurately, it is necessary to evaluate the whole-body activity of a person, which will consume a lot of time to process the data and cannot be applied in time. To solve these problems, in this paper we propose RF-Detection, a person detection system using RFID. First of all, RF-Detection takes step length as the standard for person detection, divides step length into specific sections according to the relationship between step length and height, and achieves high accuracy for new user detection through a large amount of training for a specific step length. Secondly, RF-Detection can better identify the number of people in the same space by segmenting continuous people. Finally, the data collection was reduced by expanding the data set, and the deep learning method was used to further improve the accuracy. The results show that the overall recognition accuracy of RF-Detection is 98.93%.

show abstract

“…The advantages of CNNs for WiFi sensing consist of fewer training parameters and the preservation of the subcarrier and time dimension in CSI data. However, the disadvantage is that CNN has an insufficient receptive field due to the limited kernel size and thus fails to capture the [40] People Counting MLP Intel 5300 NIC Supervised learning EI [41] Human Activity Recognition CNN Intel 5300 NIC Transfer learning CrossSense [29] Human Identification,Gesture Recognition MLP Intel 5300 NIC Transfer Ensemble learning [42] Human Activity Recognition LSTM Intel 5300 NIC Supervised learning DeepSense [5] Human Activity Recognition CNN-LSTM Atheros CSI Tool Supervised learning WiADG [25] Gesture Recognition CNN Atheros CSI Tool Transfer learning WiSDAR [43] Human Activity Recognition CNN-LSTM Intel 5300 NIC Supervised learning WiVi [7] Human Activity Recognition CNN Atheros CSI Tool Supervised learning SiaNet [9] Gesture Recognition CNN-LSTM Atheros CSI Tool Few-Shot learning CSIGAN [44] Gesture Recognition CNN, GAN Atheros CSI Tool Semi-Supervised learning DeepMV [45] Human Activity Recognition CNN (Attention) Intel 5300 NIC Supervised learning WIHF [46] Gesture Recognition CNN-GRU Intel 5300 NIC Supervised learning DeepSeg [47] Human Activity Recognition CNN Intel 5300 NIC Supervised learning [48] Human Activity Recognition CNN-LSTM Intel 5300 NIC Supervised learning [35] Human Activity Recognition LSTM Nexmon CSI Tool Supervised learning [49] Human Activity Recognition CNN Nexmon CSI Tool Supervised learning [50] Human Activity Recognition CNN Intel 5300 NIC Few-Shot learning Widar [31] Human Identification, Gesture Recognition CNN-GRU Intel 5300 NIC Supervised learning WiONE [51] Human Identification CNN Intel 5300 NIC Few-Shot learning [52] Human Activity Recognition CNN, RNN, LSTM Intel 5300 NIC Supervised learning THAT [53] Human Activity Recognition Transformers Intel 5300 NIC Supervised learning WiGr [54] Gesture Recognition CNN-LSTM Intel 5300 NIC Supervised learning MCBAR [55] Human Activity Recognition CNN, GAN Atheros CSI Tool Semi-Supervised learning CAUTION [12] Human Identification CNN Atheros CSI Tool Few-Shot learning CTS-AM [56] Human Activity Recognition CNN (Attention) Intel 5300 NIC Supervised learning WiGRUNT…”

Section: B Convolutional Neural Networkmentioning

confidence: 99%

Deep Learning and Its Applications to WiFi Human Sensing: A Benchmark and A Tutorial

Yang¹,

Chen²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

WiFi sensing has been evolving rapidly in recent years. Empowered by propagation models and deep learning methods, many challenging applications are realized such as WiFi-based human activity recognition and gesture recognition. However, in contrast to deep learning for visual recognition and natural language processing, no sufficiently comprehensive public benchmark exists. In this paper, we highlight the recent progress on deep learning enabled WiFi sensing, and then propose a benchmark, SenseFi, to study the effectiveness of various deep learning models for WiFi sensing. These advanced models are compared in terms of distinct sensing tasks, WiFi platforms, recognition accuracy, model size, computational complexity, feature transferability, and adaptability of unsupervised learning. It is also regarded as a tutorial for deep learning based WiFi sensing, starting from CSI hardware platform to sensing algorithms. The extensive experiments provide us with experiences in deep model design, learning strategy skills and training techniques for real-world applications. To the best of our knowledge, this is the first benchmark with an open-source library for deep learning in WiFi sensing research. The benchmark codes are available at https://github.com/CHENXINYANsg/WiFi-CSI-Sensing-Benchmark.

show abstract

DeepSeg: Deep-Learning-Based Activity Segmentation Framework for Activity Recognition Using WiFi

Cited by 55 publications

References 56 publications

RF-Eletter: A Cross-Domain English Letter Recognition System Based on RFID

RF-Eletter: A Cross-Domain English Letter Recognition System Based on RFID

Capturing Features and Performing Human Detection from Human Gaits Using RFID

Deep Learning and Its Applications to WiFi Human Sensing: A Benchmark and A Tutorial

Contact Info

Product

Resources

About