This paper proposes a human activity recognition (HAR) method for frequency-modulated continuous wave (FMCW) radar sensors. The method utilizes a multi-domain feature attention fusion network (MFAFN) model that addresses the limitation of relying on a single range or velocity feature to describe human activity. Specifically, the network fuses time-Doppler (TD) and time-range (TR) maps of human activities, resulting in a more comprehensive representation of the activities being performed. In the feature fusion phase, the multi-feature attention fusion module (MAFM) combines features of different depth levels by introducing a channel attention mechanism. Additionally, a multi-classification focus loss (MFL) function is applied to classify confusable samples. The experimental results demonstrate that the proposed method achieves 97.58% recognition accuracy on the dataset provided by the University of Glasgow, UK. Compared to existing HAR methods for the same dataset, the proposed method showed an improvement of about 0.9–5.5%, especially in the classification of confusable activities, showing an improvement of up to 18.33%.

Image colorization has a widespread application in video and image restoration in the past few years. Recently, automatic colorization methods based on deep learning have shown impressive performance. However, these methods map grayscale image input into multi-channel output directly. In the process, it usually loses detailed information during feature extraction, resulting in abnormal colors in local areas of the colorization image. To overcome abnormal colors and improve colorization quality, we propose a novel Double-Channel Guided Generative Adversarial Network (DCGGAN). It includes two modules: a reference component matching module and a double-channel guided colorization module. The reference component matching module is introduced to select suitable reference color components as auxiliary information of the input. The double-channel guided colorization module is designed to learn the mapping relationship from the grayscale to each color channel with the assistance of reference color components. Experimental results show that the proposed DCGGAN outperforms existing methods on different quality metrics and achieves state-of-the-art performance.

In recent years, the increase of multimodal image data has offered a broader prospect for multimodal semantic segmentation. However, the data heterogeneity between different modalities make it difficult to leverage complementary information and create semantic understanding deviations, which limits the fusion quality and segmentation accuracy. To overcome these challenges, we propose a hybrid attention driven CNN-Mamba multimodal fusion network (HACMNet) for semantic segmentation. It aims to fully exploit the strengths of optical images in texture and semantic representation, along with the complementary structural and elevation information from the digital surface model (DSM). This enables the effective extraction and combination of global and local complementary information to achieve higher accuracy and robustness in semantic segmentation. Specifically, we propose a progressive cross-modal feature interaction mechanism (PCMFI) in the encoder. It integrates the fine-grained textures and semantic information of optical images with the structural boundaries and spatial information of DSM, thereby facilitating more precise cross-modal feature interaction. Secondly, we design an adaptive dual-stream Mamba cross-modal fusion module (ADMCF), which leverages a learnable variable mechanism to deeply represent global semantic and spatial structural information. This enhances deep semantic feature interaction and improves the ability of model to distinguish complex land cover categories. Together, these modules progressively refine cross-modal cues and strengthen semantic interactions, enabling more coherent and discriminative multimodal fusion. Finally, we introduce a global-local feature decoder to effectively integrate the global and local information from the fused multimodal features. It preserves the structural integrity of target objects while enhancing edge detail representation, thus enhancing segmentation results. Through rigorous testing on standard datasets like ISPRS Vaihingen and Potsdam, the proposed HACMNet demonstrates advantages over prevailing methods in multimodal remote sensing analysis, particularly on challenging object classes.

Sketch face recognition has recently gained significant attention in the field of computer vision due to its ability to quickly identify matched pairs of optical and sketch images. This technology has the potential to greatly improve the efficiency of law enforcement agencies in criminal investigations. However, there are still challenges that need to be addressed in sketch face recognition algorithms, such as modal differences and limited sample sizes. To overcome these issues, this study proposes a Residual Serialized Cross Grouping Transformer (RSCGT), which contains a residual serialized module to reduce the computation complexity, a two-layer Cross Grouping Transformer module that is capable of extracting modality-invariant context features, a domain adaptive module to mitigate the impact of modal differences. Additionally, we introduce a meta-learning training strategy to augment the generalization ability of this model. Experimental results demonstrate that the RSCGT achieves high accuracy in sketch face recognition tasks, even with small-scale datasets.

Multiclass geospatial object detection in high-spatial-resolution remote-sensing images (HSRIs) has recently attracted considerable attention in many remote-sensing applications as a fundamental task. However, the complexity and uncertainty of spatial distribution among multiclass geospatial objects are still huge challenges for object detection in HSRIs. Most current remote-sensing object-detection approaches fall back on deep convolutional neural networks (CNNs). Nevertheless, most existing methods only focus on mining visual characteristics and lose sight of spatial or semantic relation discriminations, eventually degrading object-detection performance in HSRIs. To tackle these challenges, we propose a novel hybrid attention-driven multistream hierarchical graph embedding network (HA-MHGEN) to explore complementary spatial and semantic patterns for improving remote-sensing object-detection performance. Specifically, we first constructed hierarchical spatial graphs for multiscale spatial relation representation. Then, semantic graphs were also constructed by integrating them with the word embedding of object category labels on graph nodes. Afterwards, we developed a self-attention-aware multiscale graph convolutional network (GCN) to derive stronger for intra- and interobject hierarchical spatial relations and contextual semantic relations, respectively. These two relation networks were followed by a novel cross-attention-driven spatial- and semantic-feature fusion module that utilizes a multihead attention mechanism to learn associations between diverse spatial and semantic correlations, and guide them to endowing a more powerful discrimination ability. With the collaborative learning of the three relation networks, the proposed HA-MHGEN enables grasping explicit and implicit relations from spatial and semantic patterns, and boosts multiclass object-detection performance in HRSIs. Comprehensive and extensive experimental evaluation results on three benchmarks, namely, DOTA, DIOR, and NWPU VHR-10, demonstrate the effectiveness and superiority of our proposed method compared with that of other advanced remote-sensing object-detection methods.