Incremental few-shot semantic segmentation (iFSS) aims to incrementally acquire new knowledge from limited labeled samples while retaining previously learned concepts, without relying on large-scale manual annotations. Despite its practical importance, effective iFSS methods remain limited. In this paper, we propose a multimodal framework to address this challenge. First, textual features are leveraged to supervise the classifier weights, mitigating forgetting of base classes and overfitting to novel classes. Second, class prototype images are generated from textual features to support low-cost replay of previous knowledge. Finally, a momentum-based updating strategy is introduced to decouple the background and novel class weights within the classifier of the old model used for knowledge distillation. Extensive experiments and ablation studies validate the effectiveness of the approach. The method achieves strong performance in continual learning of novel classes while preserving knowledge of old ones, closely mirroring human-like few-shot learning over time.

Non-Intrusive Load Monitoring (NILM) has attracted much attention as a promising method of identifying electrical appliances. It discriminates electrical devices based on changes in load characteristics to enable electrical device scheduling strategies for optimal energy utilization. Existing NILM methods mainly use high-frequency electrical-specific signals and have high computational and memory requirements, which are difficult to implement on resource-constrained devices. Therefore, we propose NILM-LANN, a novel lightweight neural network with an attention mechanism, in which several tricks including Convolutional Long Short-Term Memory (ConvLSTM), 1-D convolutional operations and DenseNet are comprehensively utilized and balanced to reduce the number of model parameters, deepen the network while avoiding gradient explosion or vanishing. The Squeeze and Excitation block (SE-block) is also used to capture channel-wise dependencies based on the aggregated information. The NILM-LANN model is evaluated on public datasets of UK-DALE and LIT, and a self-built CAE dataset, with a maximum classification accuracy of 99.9% and F1-Score of 99.9%, while reducing the number of model parameters by over 90% compared to other existing methods such as Alex-Net and Light-LSTM Finally, the NILM-LANN is deployed in NVIDIA Jetson Nano B01 embedded device to verily its lightweight and efficiency.

Artificial intelligence, the main enabler for the intelligence of aquaculture monitoring systems, helps to increase the efficiency in fish farming, ensure the robustness and reduce the maintenance costs. With the rapid development of AI, many researchers focus on understanding the statuses of the fish via using water quality sensors and fish motion detection. However, existing methods focus on the pattern of movement of the fish, without exploring the health statuses in a real environment. Meanwhile, the computational requirement of the model and computing hardware equipment is also need to be taken into consideration. Thus, an edge-based aquaculture monitoring system is proposed with the advantage of health statuses recognition based on AI model and real-time communication by using edge computing devices. Slowfast, the deep learning algorithm is also first implemented to detect the different statuses of fish with an accuracy of 97.06%, which outperforms the other existing methods. The proposed monitoring system has also been successfully deployed in monitoring the intensive American shad farming in the cities of Suzhou, China.

Energy harvesting act as one of the promising techniques to provide sustainable energy for self-powered sensor nodes by converting environmental energy into electricity. Energy harvested from single micro-energy source suffers from low power density and vulnerable to environmental changes, while the hybrid energy harvesting system can supply energy sustainably. However, cumulative power from various energy sources leads to a challenge of low energy conversion efficiency. In this paper, a Combined Maximum Power Point Tracking (Com-MPPT) method for hybrid energy harvesting is designed to improve the overall harvesting efficiency. By analyzing output characteristic curves under different environmental states, a mathematic model for hybrid energy harvesting system is proposed. It is found that with the change of environmental factors, multiple power peaks exist with varying voltage on the load, where traditional MPPT method for single energy source is no longer applicable. Therefore, the proposed Com-MPPT algorithm based on Search skip and linear extrapolation proposed in this paper can quickly search for the global maximum power peak in the case of multiple power peaks in hybrid energy collection. Simulation and experimental results show that the proposed algorithm can track the global maximum power point rapidly under different environmental conditions, and the tracking time is more than 50% shorter than that of the improved particle swarm optimization and flower pollination algorithm.

As the growth number of electrical appliances used in industry and household, identifying the categories of the electrical devices act as a priority to enable an intelligent management system for electricity consumption. Appliance load monitoring (ALM) is essential for energy management solutions, allowing them to obtain appliance-specific energy consumption statistics that can further be used for optimal energy utilization. However, collecting electrical features from numerous and multivariate appliances may cause a high burden in the communication network and existing identification methods do not perform well for online monitoring. Thus, this paper proposed an edge monitoring system with a multilevel wavelet decomposition network for classification of electrical appliances. The proposed model can achieve an average error of 1.4% on the UCR public dataset and an error rate of 2.7% on the self-built electrical appliance dataset with a maximum service delay of 50 milliseconds.

The U-Net-like model has been widely studied in the field of building extraction. However, most of these models are based on locally sensed Convolutional Neural Networks(CNNs) designed with symmetric structure and single feature processing, which cannot accurately identify buildings with different sizes, shapes, and colors in remote sensing images. To overcome these problems, we propose the asymmetric cascade fusion network(ACFN), based on the Vision Transformer(ViT), to design a novel asymmetric architecture to recognize buildings of different sizes and shapes by processing multi-granularity features by different means. First, the asymmetric architecture obtains multi-granularity features with global contextual information by embedding different types of attention in encoder-decoders of different sizes. This architecture can identify densely distributed and occluded buildings by semantic reasoning in remote sensing images with complex information. Second, we design a multi-branch weighted pyramid pooling module, which sets different branch weights to offset the background noise introduced in introducing global contextual information. Our ACFN significantly improves the Beijing buildings, ISPRS-Vaihingen, and LoveDA datasets.

The task of human interaction understanding involves both recognizing the action of each individual in the scene and decoding the interaction relationship among people, which is useful to a series of vision applications such as camera surveillance, video-based sports analysis and event retrieval. This paper divides the task into two problems including grouping people into clusters and assigning labels to each of them, and presents an approach to solving these problems in a joint manner. Our method does not assume the number of groups is known beforehand as this will substantially restrict its application. With the observation that the two challenges are highly correlated, the key idea is to model the pairwise interacting relations among people via a complete graph and its associated energy function such that the labeling and grouping problems are translated into the minimization of the energy function. We implement this joint framework by fusing both deep features and rich contextual cues, and learn the fusion parameters from data. An alternating search algorithm is developed in order to efficiently solve the associated inference problem. By combining the grouping and labeling results obtained with our method, we are able to achieve the semantic-level understanding of human interactions. Extensive experiments are performed to qualitatively and quantitatively evaluate the effectiveness of our approach, which outperforms state-of-the-art methods on several important benchmarks. An ablation study is also performed to verify the effectiveness of different modules within our approach.