Recognizing prohibited items automatically is of great significance for intelligent X-ray baggage security screening. Convolutional neural networks (CNNs), with the support of big training data, have been verified as the powerful models capable of reliably detecting the expected objects in images. Therefore, building a specific CNN model working reliably on prohibited item detection also requires large amounts of labeled item image data. Unfortunately, the current X-ray baggage image database is not big enough in count and diversity for CNN model training. In this paper, we propose a novel method for X-ray prohibited item data augmentation using generative adversarial networks (GANs). The prohibited items are first extracted from X-ray baggage images using a K-nearest neighbor matting scheme. Then, the poses of the obtained item images are estimated using a space rectangular coordinate system and categorized into four or eight classes for constructing a training database. For generating the realistic samples reliably, different GAN models are evaluated using Frechet Inception Distance scores, and some important tips of handling GAN training in X-ray prohibited item image generation are reported. Finally, to verify whether the generated images belong to its corresponding class or not, a cross-validation scheme based on a CNN model is implemented. The experimental results show that most of the generated images can be classified correctly by the CNN model. This implies that the generated prohibited item images can be used as the extended samples to augment an X-ray image database.

Traffic flow prediction is a critical component of intelligent transportation systems, playing a vital role in alleviating congestion, improving road resource utilization, and supporting traffic management decisions. Although deep learning methods have made remarkable progress in this field in recent years, current studies still face challenges in modeling complex spatio-temporal dependencies, adapting to anomalous events, and generalizing to large-scale real-world scenarios. To address these issues, this paper proposes a novel traffic flow prediction model. The proposed approach simultaneously leverages temporal and frequency domain information and introduces adaptive graph convolutional layers to replace traditional graph convolutions, enabling dynamic capture of traffic network structural features. Furthermore, we design a frequency–temporal multi-head attention mechanism for effective multi-scale spatio-temporal feature extraction and develop a cross-multi-scale graph fusion strategy to enhance predictive performance. Extensive experiments on real-world datasets, PeMS and Beijing, demonstrate that our method significantly outperforms state-of-the-art (SOTA) baselines. For example, on the PeMS20 dataset, our model achieves a 53.6% lower MAE, a 12.3% lower NRMSE, and a 3.2% lower MAPE than the best existing method (STFGNN). Moreover, the proposed model achieves competitive computational efficiency and inference speed, making it well-suited for practical deployment.

Sea surface temperature (SST) is a critical variable for ocean monitoring and ecosystem protection, particularly in the context of climate change. Although LSTM-based models have been widely adopted for SST prediction, they often struggle to capture localized spatial patterns effectively. To address this, we propose TS-ViT-SST, a Token Sparsification-based Vision Transformer that selectively attends to informative regions within SST data, thereby enhancing the modeling of fine-grained spatial variations. Furthermore, we introduce a physics-inspired temperature difference loss to suppress unrealistic fluctuations and improve prediction accuracy. Model performance is evaluated primarily using RMSE, with additional comparisons of parameter counts, computation times, AIC, BIC, and other relevant metrics across different models. Experiments on the SST-RSS dataset demonstrate that TS-ViT-SST surpasses state-of-the-art (SOTA) methods, achieving a 0.11 reduction in RMSE while requiring fewer computational resources. Additional validation on the ICAR-ENSO dataset for El Niño—Southern Oscillation (ENSO) prediction confirms the robustness and generalization capability of our approach. By integrating sparsified attention mechanisms with physical constraints, TS-ViT-SST establishes a new benchmark for SST forecasting. The code is available at: https://github.com/zhzhao2020/TS-ViT-SST.

Rotation of rotor blades is typically micro-motion and would induce micro-Doppler modulation. However, accurate detection will become extremely difficult when the echo signal is corrupted by strong noise. This paper proposes a method to realize extraction of micro-Doppler feature from the rotating rotor blade echo signals with low signal-to-noise ratio (SNR). The proposed method is characterized by applying low-rank decomposition for saliency detection of spectrogram. MATLAB simulation results verify that the proposed method is effective in saliency detection of the spectrogram from rotor blade echo signals and is robust in estimating the maximum micro-Doppler shift under low SNR.