Stroke is a kind of cerebrovascular disease that heavily damages people’s life and health. The quantitative analysis of brain MRI images plays an important role in the diagnosis and treatment of stroke. Deep neural networks with massive data learning ability supply a powerful tool for lesion detection. In order to study the property of the stroke lesions and complete intelligent automatic detection, we collaborated with two authoritative hospitals and collected 5,668 brain MRI images of 300 ischemic stroke patients. All the lesion regions in the images were accurately labeled by professional doctors to ensure the authority and effectiveness of the data. Three categories of deep learning object detection networks including Faster R-CNN, YOLOV3, and SSD are applied to implement automatic lesion detection with the best precision of 89.77%. Meanwhile, statistical analysis of the locations, shapes of the lesions, and possible related diseases is conducted with valid conclusions. The research contributes to the intelligent assisted diagnosis and prevention and treatment of ischemic stroke.

Different from other human behaviors, sign language has the characteristics of limited local motion of upper limb and meticulous hand action. Some sign language gestures are ambiguous in RGB video due to the influence of lighting and background color, which affects the recognition accuracy. We propose a multimodal deep learning architecture for sign language recognition which effectively combines RGB-D input and two-stream spatiotemporal networks. Depth videos, as an effective compensation of RGB input, can supply additional distance information about the signer’s hands. A novel sampling method called ARSS (Aligned Random Sampling in Segments) is put forward to select and align optimal RGB-D video frames, which improves the capacity utilization of multimodal data and reduces the redundancy. We get the hand ROI by joints information of RGB data for local focus in spatial stream. D-shift Net is proposed as depth motion feature extraction in temporal stream, which fully utilizes three dimensional motion information of the sign language. Both streams are fused by convolutional fusion layer to get complementary features. Our approach explored the multimodal information and enhanced the recognition precision. It obtains the state-the-of-art performance on the datasets of CSL (96.7%) and IsoGD (63.78%).

Semantic segmentation plays a significant role in parsing remote sensing images. However, mainstream segmentation models lack a thorough understanding of the complex structures and scale differences, and struggle to effectively locate and emphasize diverse edges. Aiming at these limitations, we propose a frequency-driven edge guidance network, named FDEG-Net, for semantic segmentation of remote sensing images. First, we design a joint sparse context aggregation module that integrates both dense local context and sparse long-range context to improve the analysis of intricate and multiscale objects. Second, an edge guidance module is developed for strong interclass edge acquisition. It applies a 2-D discrete wavelet transform, coefficient superposition method, and adaptive edge feature enhancement algorithm to reduce low-frequency information and highlight salient boundaries in spatial features. This module has two significant advantages. 1) The edge positions are defined in pixel intensity with high interpretability. 2) The modular design without additional edge labels is plug-and-play. The effectiveness and robustness of this module are validated through edge visualization results. The proposed FDEG-Net is evaluated on the Potsdam, Vaihingen, and GID datasets, demonstrating its excellent performance in accurately capturing the rich semantics of geographic space features.

Continuous sign language recognition (CSLR) is challenging due to the complexity of video background, hand gesture variability, and temporal modeling difficulties. This work proposes a CSLR method based on a spatial-temporal graph attention network to focus on essential features of video series. The method considers local details of sign language movements by taking the information on joints and bones as inputs and constructing a spatial-temporal graph to reflect inter-frame relevance and physical connections between nodes. The graph-based multi-head attention mechanism is utilized with adjacent matrix calculation for better local-feature exploration, and short-term motion correlation modeling is completed via a temporal convolutional network. We adopted BLSTM to learn the long-term dependence and connectionist temporal classification to align the word-level sequences. The proposed method achieves competitive results regarding word error rates (1.59%) on the Chinese Sign Language dataset and the mean Jaccard Index (65.78%) on the ChaLearn LAP Continuous Gesture Dataset.

Change detection (CD) is a significant branch of remote sensing image analysis. Its main distinction from general semantic segmentation lies in the input of bitemporal images. Recent CD methods have primarily focused on Euclidean space, disregarding the hidden non-Euclidean details due to the high imaging altitude and complex scenes in remote sensing imagery. This limitation hampers the model's performance. In recent years, hyperbolic space has gradually been introduced into deep learning as a classical non-Euclidean space. To explore the potential impact of hyperbolic information in CD, a hyperbolic-embedding-based Siamese network is proposed in this article. Specifically, we propose a hyperbolic similarity attention mechanism that can map deep features from bitemporal images into hyperbolic space, then establish the relationship between bitemporal features based on the hyperbolic distance, fully mining non-Euclidean information, and fusing features from both branches to enhance feature cohesiveness. Furthermore, we design a differential feature enhanced module in the decoder, which utilizes differential operations at multiple scales to highlight the essence of each layer's features and improve feature richness. Experimental results on two public very-high-resolution CD datasets demonstrate that the proposed network achieves better detection accuracy than other state-of-the-art CD methods.