No abstract

Activity recognition which aims to accurately distinguish human actions in complex environments plays a key role in human-robot/computer interaction. However, long-lasting and similar actions will cause poor feature sequence extraction and thus lead to a reduction of the recognition accuracy. We propose a novel discriminative deep model (D3D-LSTM) based on 3D-CNN and LSTM for both single-target and interaction action recognition to improve the spatiotemporal processing performance. Our models have several notable properties: 1) A real-time feature fusion method is used to obtain a more representative feature sequence through composition of local mixtures for enhancing the performance of discriminating similar actions; 2) We introduce an improved attention mechanism that focuses on each frame individually by assigning different weights in real-time; 3) An alternating optimization strategy is proposed for our model to obtain parameters with the best performance. Because the proposed D3D-LSTM model is efficient enough to be used as a detector that recognizes various activities, a Real-set database is collected to evaluate action recognition in complex real-world scenarios. For long-term relations, we update the present memory state via the weight-controlled attention module that enables the memory cell to store better long-term features. The densely connected bimodal modal makes local perceptrons of 3D-Conv motion-aware and stores better short-term features. The proposed D3D-LSTM model has been evaluated through a series of experiments on the Real-set and open-source datasets, i.e. SBU-Kinect and MSR-action-3D. Experimental results show that the proposed D3D-LSTM model achieves new state-of-the-art results, including pushing the average rate of the SBU-Kinect to 92.40% and the average rate of the MSR-action-3D to 95.40%.

No abstract

No abstract

Most East Asian rehabilitation centers offer chopsticks manipulation tests (CMT). In addition to impaired hand function, approximately two-thirds of stroke survivors have visual impairment related to eye movement. This article investigates the significance of combining finger joint angle estimation and a visual attention measurement in CMT. We present a multiscopic framework that consists of microscopic, mesoscopic, and macroscopic levels. We develop a feature extraction technique to extract the kinematic finger model at the microscopic level. At the mesoscopic level, we propose an active perception ability to detect the position and geometry of the finger on the chopsticks. The proposed framework estimates the proximal interphalangeal (PIP) joint angle on the index finger during CMT using fully connected cascade neural networks (FCC-NN). At the macroscopic level, we implement a cognitive ability by measuring visual attention during CMT. We further evaluate the proposed framework with a conventional test that counts the number of peanuts (NP) which are moved from one bowl to another using chopsticks within a particular time frame. For the evaluation, we introduce three parameters, namely joint angle estimation movement (JAEM), chopstick attention movement (CAM), and chopstick tip movement (CTM), by detecting the local minima and maxima of the signal. According to the experiment results, the velocity of these three parameters could indicate improvement in hand and eye function during CMT. We expect this study to benefit therapists and researchers by providing valuable information that is not accessible in the clinic. Code and datasets are available online at https://github.com/anom-tmu/cmt-attention.INDEX TERMS hand-eye interaction, rehabilitation evaluation, first-person vision, eye-gaze, eye-tracking.

No abstract

No abstract

No abstract