Gesture recognition is one of the important ways of human‐computer interaction, which is mainly detected by visual technology. The temporal and spatial features are extracted by convolution of the video containing gesture. However, compared with the convolution calculation of a single image, multiframe image of dynamic gestures has more computation, more complex feature extraction, and more network parameters, which affects the recognition efficiency and real‐time performance of the model. To solve above problems, a dynamic gesture recognition model based on CBAM‐C3D is proposed. Key frame extraction technology, multimodal joint training, and network optimization with BN layer are used for making the network performance better. The experiments show that the recognition accuracy of the proposed 3D convolutional neural network combined with attention mechanism reaches 72.4% on EgoGesture dataset, which is improved greatly compared with the current main dynamic gesture recognition methods, and the effectiveness of the proposed algorithm is verified.
Background: Previous evidence suggests that postural control processing may be more related to spatial working memory (SWM) than to nonspatial working memory (NWM). Methodological discrepancies between spatial and nonspatial cognitive tasks have made direct comparisons between the two systems difficult. Methods: To explore the neural mechanisms of SWM and NWM relative to that of postural control, participants were subjected a cognitive-posture dual-task paradigm, consisting of a 3-back letter working memory (WM) task, using physically identical stimuli with spatial and nonspatial components memorized in different sessions, and a standing balance task with a tandem stance. Additionally, there were two control sessions: a single-postural control session wherein participants pressed mouse buttons at random while standing; and a single-cognitive task control session wherein subjects completed a WM task while seated. The subjects underwent functional near-infrared spectral imaging (fNIRS) during task performance, wherein oxygenated hemoglobin concentration ([HbO]) was measured in frontal and parietal regions. Results: Postural control reduced discernment in the SWM task significantly, but did not affect NWM task performance. fNIRS showed that postural control had a significant tendency to decrease the [HbO] in the frontal-parietal network of the left hemisphere when participants completed the SWM task. No posture-associated differences in [HbO] were observed in NWM-related areas during NWM task performance. Behavioral and fNIRS data demonstrated that postural control had a selective interaction with SWM. Specifically, postural control reduced SWM discrimination and SWM-related brain activity (frontal-parietal network), but not NWM discrimination or NWM-related brain activity. Furthermore, the multiple linear regression analysis showed that SWM, but not NWM, was an important predictor of postural control. These results suggest that postural control may share more cognitive resources with SWM than with NWM.
Tire tread pattern image classification plays an important role in crime scene and traffic accident investigation. Due to the lack of standard test dataset, there is little work done in this area. For efficient texture feature description, inherent characteristic of tire patterns need to be considered. Leveraging on the directionality characteristics of tread patterns, a novel texture feature extraction algorithm is proposed based on adaptive weighted feature fusion with the weights defined by sub-band energy ratio. The proposed approach consists of: (1) discrete wavelet decomposition of tire tread image to obtain low frequency, horizontal, vertical and diagonal sub-bands; (2) extraction of rotation-invariant uniform local binary pattern features from the sub-band images; (3) concatenating the tread pattern directional features, weighted by their corresponding sub-band energies. Applying SVM for tire tread pattern classification, experimental results on real-world tire tread patterns show that the proposed texture feature extraction algorithm is outperforms other prior methods.
Background: the effect of acute exercise on cognition covers almost all stages of information processing, but few studies have focused on visual awareness. Reports on the appearance of faint speed-changes in the perception of stimuli were used as an index for visual awareness. Visual awareness was assessed after exercise or rest. Aside from the detection of speed-changes, speed-change discrimination was added as an index of perception. Results: the results showed that reports on the appearance of faint speed-changes were affected by acute aerobic exercise. The d’ index was higher after exercise. The hit rate for speed-change detection was marginally significantly higher after exercise than after the sedentary test condition. Analysis of the results obtained for the discrimination task showed that discrimination speed was boosted only when subjects were aware of the speed-change. Importantly, neither false alarm rate nor response bias was affected by exercise. Conclusions: acute moderate- to vigorous-intensity aerobic exercise improved subjects’ awareness of speed changes. In addition, there was a perceptual advantage due to exercise.