Human activity recognition (HAR) plays a vital role in many applications, such as surveillance, in-home monitoring, and health care. Portable radar sensor has been increasingly used in HAR systems in combination with deep learning (DL). However, it is both difficult and time-consuming to obtain a large-scale radar dataset with reliable labels. Insufficient labeled data often limit the generalization of DL models. As a result, the performance of DL models will drop when being applied to a new scenario. In this sense, only labeling a small portion of data in the large-scale radar dataset is more feasible. In this article, we propose a semisupervised transfer learning (TL) algorithm, 'joint domain and semantic transfer learning (JDS-TL),' for radar-based HAR, which is composed of two modules: unsupervised domain adaptation (DA) and supervised semantic transfer. By employing a sparsely labeled dataset to train the HAR model, the proposed method alleviates the need of labeling a significantly large number of radar signals. We adopt a public radar micro-Doppler spectrogram dataset including six human activities to evaluate JDS-TL. Experiments show that the proposed JDS-TL is able to recognize the six activities with an average accuracy of 87.6% when there are only 10% instances labeled in the training dataset. Ablation analysis also demonstrates the efficiency of the DA and the semantic transfer modules.

@en

Radio frequency interference, which makes it difficult to produce high-quality radar spectrograms, is a major issue for micro-Doppler-based human activity recognition (HAR). In this paper, we propose a deep-learning-based method to detect and cut out the interference in spectrograms. Then, we restore the spectrograms in the cut-out region. First, a fully convolutional neural network (FCN) is employed to detect and remove the interference. Then, a coarse-to-fine generative adversarial network (GAN) is proposed to restore the part of the spectrogram that is affected by the interferences. The simulated motion capture (MOCAP) spectrograms and the measured radar spectrograms with interference are used to verify the proposed method. Experimental results from both qualitative and quantitative perspectives show that the proposed method can mitigate the interference and restore high-quality radar spectrograms. Furthermore, the comparison experiments also demonstrate the efficiency of the proposed approach.

@en

Impurity in transparent-bottled liquid is a serious production accident in the field of beverage and medicine industry. However, the existing detection systems are difficult to distinguish impurities with dynamic interference (bubbles and stains) and detect impurities located at the edge of the bottle. In order to solve the problems stated above, a new machine vision system for detecting tiny and dynamic impurities is proposed in this paper. In the system, circularity calculation, longitudinal frame-difference method, orthogonal-axis inspection and K-Nearest Neighbor (KNN) machine learning algorithm are combined together to realize the automatic and real-time detection. Experimental results demonstrate that, after completing machine learning, the weighted error of the proposed system for detecting impurities can be effectively controlled at about 0.9% even in dynamic interference environment, which is great significance to safety production in beverage and medicine industry.

@en

Conveyor belt longitudinal tear is one of the most serious problems in coal mining. Existing systems cannot realise lossless and real-time detection for longitudinal tear of conveyor belt. Currently, visual detecting systems are proposed by many researchers and are becoming the future trend. A visual recognition system based on using laser and area light sources is designed in this study, which can recognise and count abrasions, incomplete-tears, and complete-tears. The advantage of the system is to prevent longitudinal tear based on multi-feature information. In the process of detecting conditional characteristics, laser and area light sources are responsible for enhancing contrast between conditional features and conveyor belt surface, meanwhile false corner filtration and single-point feature identification method are designed for improving recognition accuracy of the system. Compared with several current systems, the designed system has a better performance on recognising complex tear characteristics of conveyor belt, thus the problem of starting warning only based on single feature can be effectively avoided.@en