Remote-sensing semantic segmentation drives by land-use monitoring, urban planning, and ecological assessment, yet progress is hampered by scarce pixel-level labels. To address this issue, we present HyperR3SNet, which is an efficient framework for remote-sensing semantic segmentation that tackles data scarcity and scale variations in overhead imagery. HyperR3SNet transfers self-supervised vision foundation models to remote sensing, providing strong feature generalisation with minimal labeled data. Building on this, a multi-scale cross-attention module is inserted into the backbone, vision transformer layer, enabling the network to extract richer features across widely varying object scales. To keep the model lightweight, a matrix-factorised task head is employed, sharply reducing parameters and computation while sustaining accuracy. HyperR3SNet is the first model to integrate a hyperbolic pixel-level loss with vision foundation model adaptation for cross-domain and low-annotation remote sensing segmentation, leveraging the exponential expansion property of hyperbolic geometry to capture latent inter-class relations and preserve structural consistency under weak supervision. Evaluated on the widely adopted remote sensing segmentation datasets (e.g., iSAID, LoveDA, Potsdam, and Vaihingen), HyperR3SNet achieves mean IoUs of 67.60 %, 55.86 %, 80.07 %, and 96.46 %, respectively, and on average surpasses a broad range of state-of-the-art methods.

Lightning often causes death, injury, and damage to various facilities and equipment. Accurately detecting the spatial location of lightning occurrence by predicting thunderstorms and lightning is of great significance. Traditional lightning detection systems detect lightning by measuring the sound, light, and electromagnetic field information radiated by lightning. These methods typically have two problems. First, the detection process of lightning signals is susceptible to electromagnetic interference. Second, the equipment cost is high and is not friendly to some lightning detection tasks only targeted at specific scenarios. In order to detect lightning more conveniently, we propose a lightning detection model based on deep learning networks. With the increase in the use of cameras in modern society, designing lightning object detection networks based on deep learning is possible. However, two problems have been found in existing practice: (1) When strong lightning meteorological phenomena occur, the lightning features in the image are covered by bright electric lights, and convolutional neural networks cannot distinguish between strong lightning scenes and strong ultraviolet scenes. (2) The performance of convolutional neural networks is often related to the model's size. The larger the model, the stronger the performance of the network. However, in practical application scenarios, computing resources are insufficient to use sufficiently large networks. In this paper, we propose a simple and effective lightning object detection network (LD‐Net) and use a foreground‐background segmentation algorithm to locate frames containing lightning in the video. After using the knowledge distillation‐based model compression method, the mAP of the lightning object detection network with a backbone net of resnet with 18‐layer (LD‐Net‐18) can reach 82.4%. We hope that the proposed LD‐Net can serve as a simple and powerful alternative to traditional lightning detection methods, enhancing efficiency in lightning detection tasks.

In multi-resolution remote sensing imagery, roads typically exhibit sparse, elongated, and structurally complex morphological characteristics, posing formidable connectivity modeling challenges for semantic segmentation models. Existing approaches predominantly focus on pixel-level accuracy, often neglecting the topological integrity of road networks, which leads to frequent discontinuities and omissions in predicted results. To address this, this paper proposes an end-to-end road extraction framework equipped with multi-receptive field modeling and structural connectivity preservation capabilities. The model incorporates a multi-receptive-field module to capture road patterns across varying spatial scales, a connectivity-aware decoding mechanism to strengthen structural coherence, and a topology-aware loss that explicitly guides the restoration of continuous road networks during training. On the DeepGlobe-Road dataset, TopoRF-Net achieves OA 98.57%, IoU 69.76%, F1-score 82.18%, Precision 85.50%, and Recall 79.12%; on the Massachusetts dataset, TopoRF-Net similarly achieved outstanding results: OA 96.65%, IoU 59.68%, F1-score 74.75%, Precision 77.98%, and Recall 71.77%. These results conclusively demonstrate that the proposed method significantly outperforms existing approaches in both precision and connectivity metrics, whilst exhibiting favorable parameter efficiency and inference performance.