To achieve device-free indoor localization without the active participation of the users, this paper presents WallSense, a device-free indoor localization system based on off-the-shelf Radio RFID (Radio-Frequency Identification) equipment. The system deploys two orthogonal tag arrays in adjoining walls and uses the RSSI and phase information measured by RFID readers to localize the target. By differentiating the backscattered signal between adjacent tag pairs, WallSense is able to eliminate most undesirable factors and extract information directly related to the location of the target. By applying Particle Swarm Optimization (PSO) with a novel weighted fitness function and combining the localization result of two orthogonal tag arrays, the system is able to localize the target with high accuracy. Experiments show that the system is able to localize human target with 0.24 m median error. Also, WallSense has low deployment overhead and do not require the users to carry any devices.

Gesture recognition, the basis of human–computer interaction (HCI), is a significant component for the development of smart home, VR, and senior care management. Most gesture recognition methods still depend on sensors worn by the user or video-based gestures for recognition, can be used for fine-grained gesture recognition. our paper implements a gesture recognition method that is independent of environment and gesture drawing direction, and it achieves gesture recognition classification by using small sample data. Wi-NN, proposed in this study, does not require the user to wear additional device. In this case, channel state information (CSI) extracted from Wi-Fi signal is used to capture the action information of the human body via CSI. After pre-processing to reduce the interference of environmental noise as much as possible, clear action information is extracted using the feature extraction method based on time domain to obtain the gesture action feature data. The gathered data are integrated with the weighted k-nearest neighbor (KNN) classification recognizer for classification task. The experiment outcomes revealed that the accuracy scores of the same gesture for different users and different gestures for the same user under the same environment were 93.1% and 89.6%, respectively. The experiments in different environments also achieved good recognition results, and by comparing with other experimental methods, the experiments in this paper have better recognition results. Evidently, good classification results were generated after the original data were processed and incorporated into the weighted KNN.

As the number of individuals with speech and hearing impairments continues to grow, the demand for sign language recognition systems is increasing. Furthermore, most radio frequency-based gesture recognition systems have primarily focused on simple gestures, neglecting more complex actions such as sign language. Therefore, this paper proposes RF-SL, a commercial RFID-based contactless sign language recognition system. This system does not require users to wear any tags and only requires them to perform sign language gestures positioned between an antenna and a fixed multi-tag array to collect sign language signals. Firstly, by removing static reflection noise, we reduce the impact of environmental noise and obtain purer data. Secondly, to accurately delineate the start and end points of sign language action signals, we propose an improved version of the Varri signal segmentation algorithm called Varri+ to effectively optimize the segmentation of sign language actions. Thirdly, we combine the CNN model, known for its robust feature extraction capabilities, with the BiLSTM model, which excels in feature fusion. Extensive testing and evaluation were conducted in two realistic scenarios. The experimental results show that RF-SL achieves an overall average recognition accuracy of 96.8%, with an average recognition accuracy of 96.3% for new users. Additionally, RF-SL achieves an F1 score of 97.5% in an empty room with the weak multipath effects and an F1 score of 93.6% in a classroom with the strong multipath effects. These results demonstrate that our system is dynamic, flexible and robust in sign language recognition.

Gesture recognition is the basis of human-computer interaction (HCI), and RFID has attracted wide attention for its advantages of lightweight, low cost, and universality. We propose RF-CGR, which implements cross-domain Chinese character gesture recognition intuitively and effectively. The fundamental concept of RF-CGR is to extract hand motion patterns from RFID phase information and display distinctive gesture features. Firstly, we build a signal sensing model to capture gesture information effectively. Secondly, we innovatively convert the Chinese character phase signal into an intuitive picture representation and use the gesture movement change pattern as input through data pre-processing operations. Finally, Modify-Visual Geometry Group Network (M-VGG) model is used for cross-domain gesture recognition. The M-VGG network learns high-level semantic features of gestures through spatiotemporal convolutions to extract cross-domain gesture features. Compared with other models, M-VGG has fewer parameters, and shorter training time, while still achieving high-accuracy gesture recognition. We evaluate its performance in three scenarios. The accuracy is 98.75% and 98% for new users and new scenarios respectively, significantly outperforming existing wireless signal-based gesture recognition methods.

Varieties of raisins are diverse, and different varieties have different nutritional properties and commercial value. In this paper, we propose a method to identify different varieties of raisins by combining near-infrared (NIR) spectroscopy and machine learning algorithms. The direct averaging of the spectra taken for each sample may reduce the experimental data and affect the extraction of spectral features, thus limiting the classification results, due to the different substances of grape skins and flesh. Therefore, this experiment proposes a method to fuse the spectral features of pulp and peel. In this experiment, principal component analysis (PCA) was used to extract baseline corrected features, and linear models of k-nearest neighbor (KNN) and linear discriminant analysis (LDA) and nonlinear models of back propagation (BP), support vector machine with genetic algorithm (GA-SVM), grid search-support vector machine (GS-SVM) and particle swarm optimization with support vector machine (PSO- SVM) coupling were used to classify. This paper compared the results of four experiments using only skin spectrum, only flesh spectrum, average spectrum of skin and flesh, and their spectral feature fusion. The experimental results showed that the accuracy and Macro-F1 score after spectral feature fusion were higher than the other three experiments, and GS-SVM had the highest accuracy and Macro-F1 score of 94.44%. The results showed that feature fusion can improve the performance of both linear and nonlinear models. This may provide a new strategy for acquiring spectral data and improving model performance in the future. The code is available at https://github.com/L-ain/Source.