Search

The health of the trees in the forest affects the ecological environment, so timely detection of Standing Dead Trees (SDTs) plays an important role in forest management. However, due to the large spatial scope of forests, it is difficult to find SDTs through conventional approaches such as field inventories. In recent years, the development of deep learning and Unmanned Aerial Vehicle (UAV) has provided technical support for low-cost real-time monitoring of SDTs, but the inability to fully utilize global features and the difficulty of small-scale SDTs detection have brought challenges to the detection of SDTs in visible light images. Therefore, this paper proposes a multi-scale attention mechanism detection method for identifying SDTs in UAV RGB images. This method takes Faster-RCNN as the basic framework and uses Swin-Transformer as the backbone network for feature extraction, which can effectively obtain global information. Then, features of different scales are extracted through the feature pyramid structure and feature balance enhancement module. Finally, dynamic training is used to improve the quality of the model. The experimental results show that the algorithm proposed in this paper can effectively identify the SDTs in the visible light image of the UAV with an accuracy of 95.9%. This method of SDTs identification can not only improve the efficiency of SDTs exploration, but also help relevant departments to explore other forest species in the future. [ABSTRACT FROM AUTHOR]

Recent research advances in salient object detection (SOD) could largely be attributed to ever-stronger multi-scale feature representation empowered by the deep learning technologies. The existing SOD deep models extract multi-scale features via the off-the-shelf encoders and combine them smartly via various delicate decoders. However, the kernel sizes in this commonly-used thread are usually “fixed”. In our new experiments, we have observed that kernels of small size are preferable in scenarios containing tiny salient objects. In contrast, large kernel sizes could perform better for images with large salient objects. Inspired by this observation, we advocate the “dynamic” scale routing (as a brand-new idea) in this paper. It will result in a generic plug-in that could directly fit the existing feature backbone. This paper’s key technical innovations are two-fold. First, instead of using the vanilla convolution with fixed kernel sizes for the encoder design, we propose the dynamic pyramid convolution (DPConv), which dynamically selects the best-suited kernel sizes w.r.t. the given input. Second, we provide a self-adaptive bidirectional decoder design to accommodate the DPConv-based encoder best. The most significant highlight is its capability of routing between feature scales and their dynamic collection, making the inference process scale-aware. As a result, this paper continues to enhance the current SOTA performance. Both the code and dataset are publicly available at https://github.com/wuzhenyubuaa/DPNet. [ABSTRACT FROM AUTHOR]

Frequency response analysis (FRA) is a widely used approach for detecting winding faults in a transformer. Appropriate and quantitative FRA features will help to improve the accuracy of fault diagnosis. In this study, a novel FRA interpretation including new image features is proposed based on the image processing technique. First, winding faults of different windings are simulated in a test autotransformer and the FRA curves are measured under various faults. Then frequency region division method and image processing technique are first applied to the measured FRA curves. The area ratio and centroid deviation in different frequency regions are calculated through a novel algorithm. Finally, the image features are used as the inputs to support vector machine model. Additionally, three different parametric optimisation algorithm are compared during the training process. The results show that the particle swarm optimisation and image feature exhibit best performance for identifying winding faults. [ABSTRACT FROM AUTHOR]

The instability of wind power has a serious impact on the power grid system, and an accurate wind power forecasting is greatly demanded in practice. Due to the time-varying nature, the distribution of weather variables such as wind speed, wind direction and temperature and/or the functional relationship of the power output relating to these weather variables are likely to change over time, which are often known as "concept drifts". However, most of the prediction models usually fail to adapt to such concept drifts. This would deteriorate the prediction accuracy, especially for the short-term prediction with high accuracy requirements. Motivated by this, this paper proposes an adaptive short-term wind power prediction method, aimed at automatically detecting the occurrence of concept drifts and updating the forecast model accordingly. The proposed method consists of three main steps, extract sample-related and sequence-related features of the weather data, cluster the data based on the similarity of the features extracted, and develop a new power prediction model with automatic drift detection and adaptation for each cluster. The comparison results show that the prediction performance of the proposed method is superior to that of the competitive methods. • Develop a new method, DDA-LightGBM, for very short-term. • The new method can take weather drifts into account. • Compare DDA-LightGBM with a wide variety of prediction models. • Show that DDA-LightGBM is able to provide a better forecasting accuracy. [ABSTRACT FROM AUTHOR]