Wireless sensor network (WSNs) stands out as a burgeoning and promising domain in intelligent sensing. Owing to various factors such as sudden sensor malfunctions or deliberate shutdown of partial nodes to save energy, the collected sensing signals from WSNs commonly have massive missing data, leading to adverse effects on subsequent analysis or decision-making. Latent factor learning (LFL) has proven to be highly effective in recovering the missing data for WSNs. However, the existing LFL models require the collected sensing signals to be maintained in one central place like a central server, which is becoming unacceptable for data owners who are getting increasingly privacy-sensitive. To address this issue, this paper innovatively proposes a f ederated l atent f actor l earning (FLFL) model for privacy-preserving spatio-temporal signal recovery. Its main idea is two-fold: 1) it designs a sensor-level federated learning framework based on LFL, where each sensor only needs to upload gradient information rather than raw data for training a privacy-preserving recovery model, and 2) it incorporates the spatio-temporal correlation into the designed federated learning framework as the regularization constraint to improve its recovery accuracy. With such designs, FLFL can not only accurately recover the missing data of WSNs but also ensure data owners' privacy-preserving of raw data. To evaluate the proposed FLFL model, extensive experiments have been conducted on four real-world WSNs datasets. The results demonstrate that FLFL significantly outperforms five state-of-the-art federated signal recovery models in terms of recovery accuracy with privacy-preserving.

Forest tree species classification has great significance for sustainable development of forest resource. Multi-source remote sensing data provides abundant temporal, spatial, and spectral information for tree species classification. However, there lacks tree species classification methods which comprehensively capture and fuse spatio-temporal-spectral information. Therefore, a tree species classification method based on deep ensemble learning of multi-source spatio-temporal-spectral remote sensing data is proposed. Firstly, multi-temporal, high-resolution and hyperspectral data are utilized for training temporal, spatial, and spectral deep networks. Furtherly, deep ensemble learning is developed for fusion of spatio-temporal-spectral network outputs, where weighted fusion is implemented via dynamic weight optimization based on the spatio-temporal-spatial features. Experimental results indicate that the importance of temporal features is higher than that of spatial information, and spectral networks perform best among all network structures. After the spatio-temporal-spectral ensemble learning, the performance of tree species classification is further improved, and the overall accuracy of the proposed method reaches above 90%. The proposed algorithm realizes precise and fine-scale tree species classification, and provides technique support for the monitoring and conservation of forest resource.

Human reaction generation represents a significant research domain for interactive AI, as humans constantly interact with their surroundings. Previous works focus mainly on synthesizing the reactive motion given a human motion sequence. This paradigm limits interaction categories to human-human interactions and ignores emotions that may influence reaction generation. In this work, we propose to generate 3D human reactions from RGB videos, which involves a wider range of interaction categories and naturally provides information about expressions that may reflect the subject's emotions. To cope with this task, we present HERO, a simple yet powerful framework for Human rEaction geneRation from videOs. HERO considers both global and frame-level local representations of the video to extract the interaction intention, and then uses the extracted interaction intention to guide the synthesis of the reaction. Besides, local visual representations are continuously injected into the model to maximize the exploitation of the dynamic properties inherent in videos. Furthermore, the ViMo dataset containing paired Video-Motion data is collected to support the task. In addition to human-human interactions, these video-motion pairs also cover animal-human interactions and scene-human interactions. Extensive experiments demonstrate the superiority of our methodology. The code and dataset will be publicly available at https://jackyu6.github.io/HERO.

Wireless sensor networks (WSNs) have emerged as a promising field of development in recent years. Due to various factors such as device short circuits or intentional node shutdowns to save energy, the collected sensing data from WSNs often contains a large number of missing values. A widely used strategy for handling this problem is low-rank matrix approximation (LRMA), which enables effective reconstruction of incomplete sensing data. To overcome the limitation that most LRMA-based methods do not fully exploit the correlations among multiple sensing indicators, this article proposes a novel spatio-temporal signal recovery model based on multi-indicator latent factorization of tensors (LFT), termed MILFT-SSR. The core contributions are twofold: (1) a multi-indicator recovery framework that leverages LFT to jointly model and utilize the dependencies among different indicators, and (2) an explicit incorporation of spatial correlation into the LFT formulation as a constraint, thereby enhancing reconstruction accuracy. Extensive experiments on two real-world WSNs datasets demonstrate that the proposed MILFT-SSR model consistently surpasses six state-of-the-art approaches in terms of recovery performance.

Understanding egocentric human-object interaction (HOI) is a fundamental aspect of human-centric perception, facilitating applications like AR/VR and embodied AI. For the egocentric HOI, in addition to perceiving semantics e.g., ''what'' interaction is occurring, capturing ''where'' the interaction specifically manifests in 3D space is also crucial, which links the perception and operation. Existing methods primarily leverage observations of HOI to capture interaction regions from an exocentric view. However, incomplete observations of interacting parties in the egocentric view introduce ambiguity between visual observations and interaction contents, impairing their efficacy. From the egocentric view, humans integrate the visual cortex, cerebellum, and brain to internalize their intentions and interaction concepts of objects, allowing for the pre-formulation of interactions and making behaviors even when interaction regions are out of sight. In light of this, we propose harmonizing the visual appearance, head motion, and 3D object to excavate the object interaction concept and subject intention, jointly inferring 3D human contact and object affordance from egocentric videos. To achieve this, we present EgoChoir, which links object structures with interaction contexts inherent in appearance and head motion to reveal object affordance, further utilizing it to model human contact. Additionally, a gradient modulation is employed to adopt appropriate clues for capturing interaction regions across various egocentric scenarios. Moreover, 3D contact and affordance are annotated for egocentric videos collected from Ego-Exo4D and GIMO to support the task. Extensive experiments on them demonstrate the effectiveness and superiority of EgoChoir.

Abstract Accurate and efficient extraction of water body information from remote sensing images is of great help to monitor water resources at the macro level, natural disaster prediction, and water pollution detection and prevention. Although many large models have achieved extremely high accuracy in remote sensing image water segmentation tasks, lightweight models are still a non‐negligible choice for many application scenarios because of the limitation of computing and storage resources. Here, WaterSegformer is described, an efficient and powerful lightweight water body segmentation model based on Segformer‐b0. The Deepmask module is designed to make the model pay more attention to the details in the image and use Lovász loss to improve IoU. In addition, DeepLabv3+ is used as the teacher model to guide the training of the model in the way of relational knowledge distillation. WaterSegformer realizes 95.06% mIoU on the test set with only 6.38 G and 3.72 M of FLOPs and parameters, respectively. Experimental results show that WaterSegformer achieves an excellent balance between accuracy, computational complexity and model size, which is hardware‐friendly, easy to deploy and enables real‐time segmentation. This method provides a new idea for water body information extraction from remote sensing images in practical applications.