High Speed and Accuracy of Animation 3D Pose Recognition Based on an Improved Deep Convolution Neural Network

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(13), P. 7566 - 7566

Published: June 27, 2023

Pose recognition in character animations is an important avenue of research computer graphics. However, the current use traditional artificial intelligence algorithms to recognize animation gestures faces hurdles such as low accuracy and speed. Therefore, overcome above problems, this paper proposes a real-time 3D pose system, which includes both facial body poses, based on deep convolutional neural networks further designs single-purpose estimation system. First, we transformed human extracted from input image abstract data structure. Subsequently, generated required at runtime dataset. This challenges conventional concept monocular estimation, extremely difficult achieve. It can also achieve running speed resolution 384 fps. The proposed method was used identify multiple-character using multiple datasets (Microsoft COCO 2014, CMU Panoptic, Human3.6M, JTA). results indicated that improved algorithm performance by approximately 3.5% 8–10 times, respectively, significantly superior other classic algorithms. Furthermore, tested system pose-recognition datasets. attitude reach 24 fps with error 100 mm, considerably less than 2D 60 learning study yielded surprisingly performance, proving deep-learning technology for has great potential.

Language: Английский

---

Computer Vision in Clinical Neurology

JAMA Neurology, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 17, 2025

Importance Neurological examinations traditionally rely on visual analysis of physical clinical signs, such as tremor, ataxia, or nystagmus. Contemporary score-based assessments aim to standardize and quantify these observations, but tools suffer from clinimetric limitations often fail capture subtle yet important aspects human movement. This poses a significant roadblock more precise personalized neurological care, which increasingly focuses early stages disease. Computer vision, branch artificial intelligence, has the potential address challenges by providing objective measures signs based solely video footage. Observations Recent studies highlight computer vision measure disease severity, discover novel biomarkers, characterize therapeutic outcomes in neurology with high accuracy granularity. may enable sensitive detection movement patterns that escape eye, aligning an emerging research focus stages. However, accessibility, ethics, validation need be addressed for widespread adoption. In particular, improvements usability algorithmic robustness are key priorities future developments. Conclusions Relevance technologies have revolutionize practice objective, quantitative signs. These could enhance diagnostic accuracy, improve treatment monitoring, democratize specialized care. Clinicians should aware their complement traditional assessment methods. further focusing validation, ethical considerations, practical implementation is necessary fully realize neurology.

Language: Английский

---

A systematic review of research on personal thermal comfort using infrared technology

Energy and Buildings, Journal Year: 2023, Volume and Issue: 301, P. 113666 - 113666

Published: Oct. 21, 2023

Language: Английский

---

Vision Transformers (ViT) for Blanket-Penetrating Sleep Posture Recognition Using a Triple Ultra-Wideband (UWB) Radar System

Sensors, Journal Year: 2023, Volume and Issue: 23(5), P. 2475 - 2475

Published: Feb. 23, 2023

Sleep posture has a crucial impact on the incidence and severity of obstructive sleep apnea (OSA). Therefore, surveillance recognition postures could facilitate assessment OSA. The existing contact-based systems might interfere with sleeping, while camera-based introduce privacy concerns. Radar-based overcome these challenges, especially when individuals are covered blankets. aim this research is to develop nonobstructive multiple ultra-wideband radar system based machine learning models. We evaluated three single-radar configurations (top, side, head), dual-radar (top + top head, side one tri-radar configuration in addition models, including CNN-based networks (ResNet50, DenseNet121, EfficientNetV2) vision transformer-based (traditional transformer Swin Transformer V2). Thirty participants (n = 30) were invited perform four recumbent (supine, left side-lying, right prone). Data from eighteen randomly chosen for model training, another six participants’ data 6) validation, remaining testing. head achieved highest prediction accuracy (0.808). Future may consider application synthetic aperture technique.

Language: Английский

---

Metrological Evaluation of Contactless Sleep Position Recognition Using an Accelerometric Smart Bed and Machine Learning

Sensors and Actuators A Physical, Journal Year: 2025, Volume and Issue: unknown, P. 116309 - 116309

Published: Feb. 1, 2025

Language: Английский

---

Sleep Posture Recognition Method Based on Sparse Body Pressure Features

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(9), P. 4920 - 4920

Published: April 29, 2025

Non-visual techniques for identifying sleep postures have become essential enhancing health. Conventional methods depend on a costly professional medical apparatus that is challenging to adapt domestic use. This study developed an economical airbag mattress and introduced method detecting sleeping positions via restricted body pressure data. The methodology relies distributed data obtained from barometric sensors positioned at various locations the mattress. Two combinations of base learners were chosen based complementary attributes model, each which can be amalgamated through soft-voting strategy. Additionally, architectures Autoencoder convolutional neural networks integrated, collectively constituting learning layer model. Gradient enhancement was utilized in meta-learner amalgamate output basic layer. experimental findings indicate suggested holistic model has high classification accuracy up 95.95%, precision 96.13%, F1 index 95.01% posture recognition assessments possesses considerable merit. In subsequent application, monitoring device identified employed air conditioner purifier create more comfortable environment. user utilize improve quality prevent related diseases.

Language: Английский

---

Soft robotics and computational intelligence: Transformative technologies reshaping biomedical engineering

Elsevier eBooks, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 14

Published: Jan. 1, 2025

Language: Английский

---

Sleep posture analysis: state-of-the-art and opportunities of wearable technologies from clinical, sensing and intelligent perception perspectives

Elsevier eBooks, Journal Year: 2025, Volume and Issue: unknown, P. 109 - 134

Published: Jan. 1, 2025

Language: Английский

---

Deciphering Optimal Radar Ensemble for Advancing Sleep Posture Prediction through Multiview Convolutional Neural Network (MVCNN) Approach Using Spatial Radio Echo Map (SREM)

Sensors, Journal Year: 2024, Volume and Issue: 24(15), P. 5016 - 5016

Published: Aug. 2, 2024

Assessing sleep posture, a critical component in tests, is crucial for understanding an individual's quality and identifying potential disorders. However, monitoring posture has traditionally posed significant challenges due to factors such as low light conditions obstructions like blankets. The use of radar technolsogy could be solution. objective this study identify the optimal quantity placement sensors achieve accurate estimation. We invited 70 participants assume nine different postures under blankets varying thicknesses. This was conducted setting equipped with baseline eight radars-three positioned at headboard five along side. proposed novel technique generating maps, Spatial Radio Echo Map (SREM), designed specifically data fusion across multiple radars. Sleep estimation using Multiview Convolutional Neural Network (MVCNN), which serves overarching framework comparative evaluation various deep feature extractors, including ResNet-50, EfficientNet-50, DenseNet-121, PHResNet-50, Attention-50, Swin Transformer. Among these, DenseNet-121 achieved highest accuracy, scoring 0.534 0.804 nine-class coarse- four-class fine-grained classification, respectively. led further analysis on ensemble For radars head, single left-located proved both essential sufficient, achieving accuracy 0.809. When only one central head used, omitting side retaining three upper-body resulted accuracies 0.779 0.753, established foundation determining sensor configuration application, while also exploring trade-offs between fewer sensors.

Language: Английский

---

DeepPose: An Integrated Deep Learning Model for Posture Detection Using Image and Skeletal Data

2022 13th International Conference on Computing Communication and Networking Technologies (ICCCNT), Journal Year: 2023, Volume and Issue: unknown

Published: July 6, 2023

Identifying human actions and postures presents significant challenges for computerized systems. The categorization of these tasks holds particular relevance in the fields health robotics. Leveraging artificial intelligence technologies, it becomes feasible to define classify recurring physical movements accurately. Proper posture is further essential rehabilitation patients because affects effectiveness exercise training. Unfortunately, fail follow correct sequence when performing exercises. To pursue problem, a new method proposed recognition pose estimation that does not require wearable devices. model utilizes 2D coordinates derived from poses as inputs with 18 joints body key points, along an image dataset, accurately various postures. This study involved training custom CNN named DeepPose using both keypoint datasets conducting comparative analysis performance two other pre-trained models. result shows dataset outperforms over dataset.

Language: Английский

---