Real-time Sleeping Posture Recognition For Smart Hospital Beds

Doan, Ngoc Phu; Pham, N.; Pham, Manh Hung; Nguyen, Huu Trung; Nguyễn, Thùy Anh; Nguyễn, Huy Hoàng

doi:10.1109/mapr53640.2021.9585289

Cited by 11 publications

(6 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, several other loss functions were mixed with the Cross-Entropy loss (CE) function: Circle loss (Cir), Contrastive loss (Con), Center loss (Cen), Triplet loss (Trip) and Triplet Center loss (Trip-Cen). Details of the training procedure including the optimizer, learning rate, step learning rate scheduler, decay rate, and batch size were the same as those employed in the work of Doan et al [6].…”

Section: Methodsmentioning

confidence: 99%

“…To enhance the accuracy of the proposed classification method, we implemented the same image preprocessing technique that was employed in the work of Doan et al [6] before feeding the pressure data image into the classifier. Since the range of raw data is from 0 to 10000, we firstly normalized the raw data to the range of the 1-channel image.…”

Section: Preprocessing Techniquementioning

confidence: 99%

“…In the inference stage, the CE-Con loss computation is removed; hence this timing difference is caused by the addition of the SPP module. According to Doan et al [6], their solution satisfies the real-time requirement with the data sampling rate of 0.3 Hz, 1 Hz, 2 Hz and 4 Hz. Therefore, our proposed solution also passes the realtime requirement at these data sampling frequencies.…”

Section: Inference Time On Various Devicesmentioning

confidence: 99%

“…The research showed 99.9 % and 87 % accuracy with 3 and 17 classes, respectively. This result was improved by the work of Doan et al [6]. In [6], the raw pressure sensor data was normalized to the range of image data, and then the obtained data were smoothed and denoised by a preprocessing technique.…”

Section: Introductionmentioning

confidence: 99%

“…This result was improved by the work of Doan et al [6]. In [6], the raw pressure sensor data was normalized to the range of image data, and then the obtained data were smoothed and denoised by a preprocessing technique. It can be seen from the above analysis that none of the studies evaluates the impact of various combinations of backbones and loss functions on the performance of sleeping posture classification.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Untitled

2022

JST: SSAD

View full text Add to dashboard Cite

Many chronic diseases, such as cardiovascular disease and sleep disorders, can be diagnosed and treated by assessing the sleeping quality. Sleep posture recognition is an important part of determining sleep quality in sleep research. While most recent studies focus on the classification problem with the number of sleeping postures being equal to or less than ten, this study aims to achieve state-of-the-art results on 17 in-bed position classification. To do this, a spatial pyramid pooling module was added to the top of the EfficientNet B0 model and the contrastive loss and cross-entropy loss functions are combined to act as the main loss function. The random 10% salt and pepper noise are utilized to augment the training data. Experimental results confirm that the proposed approach achieves the best accuracy of 96.01% and outperforms the existing methods. Additionally, we also provide an estimation related to the impact of various combinations of backbone models and loss functions on the performance of the classifier.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Preprocessing Techniquementioning

confidence: 99%

Section: Inference Time On Various Devicesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations