Search

The accurate estimation of a 3D human pose is of great importance in many fields, such as human–computer interaction, motion recognition and automatic driving. In view of the difficulty of obtaining 3D ground truth labels for a dataset of 3D pose estimation techniques, we take 2D images as the research object in this paper, and propose a self-supervised 3D pose estimation model called Pose ResNet. ResNet50 is used as the basic network for extract features. First, a convolutional block attention module (CBAM) was introduced to refine selection of significant pixels. Then, a waterfall atrous spatial pooling (WASP) module is used to capture multi-scale contextual information from the extracted features to increase the receptive field. Finally, the features are input into a deconvolution network to acquire the volume heat map, which is later processed by a soft argmax function to obtain the coordinates of the joints. In addition to the two learning strategies of transfer learning and synthetic occlusion, a self-supervised training method is also used in this model, in which the 3D labels are constructed by the epipolar geometry transformation to supervise the training of the network. Without the need for 3D ground truths for the dataset, accurate estimation of the 3D human pose can be realized from a single 2D image. The results show that the mean per joint position error (MPJPE) is 74.6 mm without the need for 3D ground truth labels. Compared with other approaches, the proposed method achieves better results. [ABSTRACT FROM AUTHOR]

To address the dangerous driving behaviors prevalent among current car drivers, it is necessary to provide real-time, accurate warning and correction of driver's driving behaviors in a small, movable, and enclosed space. In this paper, we propose a method for detecting dangerous behaviors based on frequency-modulated continuous-wave radar (mm-DSF). The highly packaged millimeter-wave radar chip has good in-vehicle emotion recognition capability. The acquired millimeter-wave differential frequency signal is Fourier-transformed to obtain the intermediate frequency signal. The physiological decomposition of the local micro-Doppler feature spectrum of the target action is then used as the eigenvalue. Matrix signal intensity and clutter filtering are performed by analyzing the signal echo model of the input channel. The signal classification is based on the estimation and variety of the feature vectors of the target key actions using a modified and optimized level fusion method of the SlowFast dual-channel network. Nine typical risky driving behaviors were set up by the Dula Hazard Questionnaire and TEIQue-SF, and the accuracy of the classification results of the self-built dataset was analyzed to verify the high robustness of the method. The recognition accuracy of this method increased by 1.97% compared with the traditional method. [ABSTRACT FROM AUTHOR]

Multimodal research is an emerging field of artificial intelligence, and the analysis of dangerous driving behavior is one of the main application scenarios in the field of multimodal fusion. Aiming at the problem of data heterogeneity in the process of behavior classification by multimodal fusion, this paper proposes a low-rank multimodal data fusion method, which utilizes the complementarity between data modalities of different dimensions in order to classify and identify dangerous driving behaviors. This method uses tensor difference matrix data to force low-rank fusion representation, improves the verification efficiency of dangerous driving behaviors through multi-level abstract tensor representation, and solves the problem of output data complexity. A recurrent network based on the attention mechanism, AR-GRU, updates the network input parameter state and learns the weight parameters through its gated structure. This model improves the dynamic connection between modalities on heterogeneous threads and reduces computational complexity. Under low-rank conditions, it can quickly and accurately classify and identify dangerous driving behaviors and give early warnings. Through a large number of experiments, the accuracy of this method is improved by an average of 1.76% compared with the BiLSTM method and the BiGRU-IAAN method in the training and verification of the self-built dataset. [ABSTRACT FROM AUTHOR]

The positioning of indoor electronic devices is an essential part of human–computer interaction, and the accuracy of positioning affects the level of user experience. Most existing methods for RF-based device localization choose to ignore or remove the impact of multipath effects. However, exploiting the multipath effect caused by the complex indoor environment helps to improve the model's localization accuracy. In response to this question, this paper proposes a multipath-assisted localization (MAL) model based on millimeter-wave radar to achieve the localization of indoor electronic devices. The model fully considers the help of the multipath effect when describing the characteristics of the reflected signal and precisely locates the target position by using the MAL area formed by the reflected signal. At the same time, for the situation where the radar in the traditional Single-Input Single-Output (SISO) mode cannot obtain the 3D spatial position information of the target, the advantage of the MAL model is that the 3D information of the target can be obtained after the mining process of the multipath effect. Furthermore, based on the original hardware, it can achieve a breakthrough in angular resolution. Experiments show that our proposed MAL model enables the millimeter-wave multipath positioning model to achieve a 3D positioning error within 15 cm. [ABSTRACT FROM AUTHOR]