Deep learning has achieved significant success in the field of hyperspectral image (HSI) classification, but challenges are still faced when the number of training samples is small. Feature fusing approaches based on multi-channel and multi-scale feature extractions are attractive for HSI classification where few samples are available. In this paper, based on feature fusion, we proposed a simple yet effective CNN-based Dual-channel Spectral Enhancement Network (DSEN) to fully exploit the features of the small labeled HSI samples for HSI classification. We worked with the observation that, in many HSI classification models, most of the incorrectly classified pixels of HSI are at the border of different classes, which is caused by feature obfuscation. Hence, in DSEN, we specially designed a spectral feature extraction channel to enhance the spectral feature representation of the specific pixel. Moreover, a spatial–spectral channel was designed using small convolution kernels to extract the spatial–spectral features of HSI. By adjusting the fusion proportion of the features extracted from the two channels, the expression of spectral features was enhanced in terms of the fused features for better HSI classification. The experimental results demonstrated that the overall accuracy (OA) of HSI classification using the proposed DSEN reached 69.47%, 80.54%, and 93.24% when only five training samples for each class were selected from the Indian Pines (IP), University of Pavia (UP), and Salinas Scene (SA) datasets, respectively. The performance improved when the number of training samples increased. Compared with several related methods, DSEN demonstrated superior performance in HSI classification.
28 January 2026 in IEEE Transactions on Geoscience and Remote Sensing
Infrared small target detection (IRSTD) is crucial for remote sensing applications like disaster warning and maritime surveillance. However, due to the lack of distinctive texture and morphological features, infrared small targets are highly susceptible to blending into cluttered and noisy backgrounds. Existing methods often rely on fixed gradient operators (e.g., Sobel, Canny) or simplistic attention mechanisms, which are inadequate for accurately extracting target edges under low contrast and high noise. In this paper, we propose an enhanced dual-path edge network (DENet) that explicitly addresses this challenge by decoupling edge enhancement and semantic modeling into two deliberately designed processing paths. The first path employs a Bidirectional-Interaction Module (BIM), which uses both Local Self-Attention and Global Self-Attention to capture multi-scale local and global feature dependencies. The global attention mechanism, based on a Transformer architecture, integrates long-range semantic relationships and contextual information, ensuring robust scene understanding. The second path introduces the Multi-Edge Refiner (Multi-ER), which enhances fine-grained edge details through multi-scale cascaded refinement. Coupled with attention-driven gating, it improves edge localization for targets of varying sizes and suppresses noise effectively. Extensive experiments on the IRSTD-1K, NUDT-SIRST and NUAA-SIRST benchmarks demonstrate that DENet significantly outperforms state-of-the-art methods, achieving superior Mean Intersection over Union and pixel-level accuracy, while maintaining lower false alarm rates. The proposed framework effectively improves infrared small target detection and localization through joint semantic modeling and edge refinement.
11 April 2025 in Proceedings of the AAAI Conference on Artificial Intelligence
Multi-object tracking faces a major challenge in handling the variations of tracked targets within complex scenes. In existing transformer-based tracking methods, typically each tracked target is only associated with one track query. However, trajectories in crowded scenes often experience varying levels of occlusion, making the association brittle for using a single track query to identify the tracked target. Therefore, we argue that relying on a single track query to track a target in complex scenes is inadequate. In this paper, we introduce TGFormer, with the core idea of designing a Track Query Group for each tracked target. Each group encompasses track queries that handle the same tracked target across different levels of occlusion scenes. To achieve long-term robust association, we propose a novel updater that integrates temporal memories and occlusion-aware features to update the Track Query Group, ensuring the tracked target can be consistently captured in complex scenes. Additionally, we introduce a Position Predictor that allows TGFormer to forecast motion trends, helping the model accurately locate moving tracklets. Experimental results show that our method achieves competitive performance on the MOT Challenge and DanceTrack datasets.
Precise monitoring of Insulated Gate Bipolar Transistors (IGBTs) junction temperature, which impacts device lifespan, is cru cial for assessing their health and operational status. However, traditional monitoring methods are limited, and effective IGBT state assessment methods are lacking. This paper presents a novel IGBT junction temperature monitoring and stat e identification method leveraging the synergy between Arrayed Waveguide Grating (AWG) demodulation and Neural Networks (NN). We first develop a comprehensive thermal simulation model of the IGBT to investigate temperature distribution, thermal stress, and potential weak p oints under diverse operating conditions. The junction temperature is calculated from the reflected wavelength data obtained by Waveguide Bragg Grating (WBG) sensors, while AWG enables high-precision, multi-channel temperature monitoring through wavelength demodulation. For stat e identification, an Attention-enhanced Siamese Network (Siam-MMSA) is proposed. By design ing a Spatial Multi-scale Attention Module (MMSA), integrating microscopic and macroscopic perspectives, this model leverages the attention mec hanism to extract relationships between temperature features and aging status. Experimental results demonstrate that this method accurately captures dynamic junction temperature variations with an error margin below 0.6°C and achieves a 99.5 % accuracy rate in IGBT stat e identification under testing conditions. This approach supports intelligent control, performance assessment, and lifespan optimization for power electronic devices, offering valuable engineering insights for real-world applications.