Search

Hand pose estimation is a fundamental task in computer vision with applications in virtual reality, gesture recognition, autonomous driving, and virtual surgery. Keypoint detection often relies on deep learning methods and high-resolution feature map representations to achieve accurate detection. The HRNet framework serves as the basis, but it presents challenges in terms of extensive parameter count and demanding computational complexity due to high-resolution representations. To mitigate these challenges, we propose a lightweight keypoint detection network called LGCANet (Lightweight Ghost-Coordinate Attention Network). This network primarily consists of a lightweight feature extraction head for initial feature extraction and multiple lightweight foundational network modules called GCAblocks. GCAblocks introduce linear transformations to generate redundant feature maps while concurrently considering inter-channel relationships and long-range positional information using a coordinate attention mechanism. Validation on the RHD dataset and the COCO-WholeBody-Hand dataset shows that LGCANet reduces the number of parameters by 65.9% and GFLOPs by 72.6% while preserving the accuracy and improves the detection speed. [ABSTRACT FROM AUTHOR]

With the increasing prominence of environmental and energy issues, electric vehicles (EVs) as representatives of clean energy vehicles have experienced rapid development in recent years, and the charging load has also exhibited statistical characteristics. Accurate prediction of EV charging load is crucial to improve grid load dispatch and intelligent level. However, current research on EV charging load prediction still faces challenges such as data reliability, complexity and variability of charging behavior, uncertainty, and lack of standardization methods. Therefore, this paper proposes an electric vehicle charging load prediction method based on spectral clustering and deep learning network (SC-CNNLSTM). Firstly, to address the insufficient amount of EV charging load data, this paper proposes to use Monte Carlo simulation to sample and simulate historical load data. Then, in order to identify the internal structure and patterns of charging load, the sampled and simulated dataset is clustered using spectral clustering, dividing the data into different clusters, where each cluster represents samples with similar charging load characteristics. Finally, based on the different sample features of each cluster, corresponding CNN-LSTM models are constructed and trained and predict using the respective data. By modifying the model parameters, the prediction accuracy of the model is improved. Through comparative experiments, the proposed method in this paper has significantly improved prediction accuracy compared to traditional prediction methods without clustering, thus validating the effectiveness and practicality of the method. [ABSTRACT FROM AUTHOR]

There are an estimated 800,000 small reservoirs globally with a range of uses. Given the collective importance of these reservoirs to water resource management and wider society, it is essential that we can monitor and understand the hydrological dynamics of ungauged reservoirs, particularly in a changing climate. However, unlike large reservoirs, continuous and systematic hydrological observations of small reservoirs are often unavailable. In response, this study has developed a retrieval framework for water levels of small reservoirs using a deep learning algorithm and remotely sensed satellite data. Demonstrated at four reservoirs in California, satellite imagery from both Sentinel-1 and Sentinel-2 along with corresponding water level field measurements was collected. Post-processed images were fed into a water level inversion convolutional neural network model for water level inversion, while different combinations of these satellite images, sampling approaches for training/testing data, and attention modules were used to train the model and evaluated for accuracy. The results show that random sampling of training data coupled with Sentinel-2 satellite imagery was generally the most accurate initially. Performance is improved by incorporating a channel attention mechanism, with the average R2 increasing by 8.6% and the average RMSE and MAE decreasing by 15.5% and 36.4%, respectively. The proposed framework was further validated on three additional reservoirs in different regions. In conclusion, the retrieval framework proposed in this study provides a stable and accurate methodology for water level estimation of small reservoirs and can be a powerful tool for small reservoir monitoring over large spatial scales. [ABSTRACT FROM AUTHOR]

Since deep learning was introduced into super-resolution (SR), SR has achieved remarkable performance improvements. Since high-level features are more informative for reconstruction, most of SR methods have a lage number of parameters, which restrict their application in resource-constrained devices. Feedback mechanism makes it possible to get informative high-level features with few parameters, for it can feed high-level features back to refine low-level ones, which is very suitable for lightweight networks. However, most feedback networks work in a single feedback manner, which refined low-level features just once in each iteration or each unit. In this paper, we propose a lightweight parallel feedback network for image super-resolution (LPFN), which enhances the refinement ability of the feedback network. In our method, all the feedback blocks feed back their outputs to previous layers in a parallel feedback manner. Based on parallel feedback and residual learning, a local-mirror architecture is proposed. Then, we propose a dispersion-aware attention residual block (DARB) as the basic block in feedback block, which calculates the dispersion of pixels along channel and spatial dimensions. We use ensemble method to reconstruct SR image. Finally, we propose a global feedback, which feeds back the degradation results of SR to primal LR image, supervising the learning of LR-HR mapping function. Further experimental results demonstrate that LPFN has an outstanding performance while taking up few computing resources. [ABSTRACT FROM AUTHOR]

• A novel drought prediction hybrid deep learning model is provided. Approximate entropy to extract data features. • Introduction of SSA-VMD to extract meteorological data information and features. • Approximate entropy for imf component screening. • Spatio-temporal visualisation of drought by models such as the spatio-temporal cube. • SSA-VMD-ARIMA-BiLSTM provides the most accurate prediction results. Jilin Province, a key agricultural hub in Northeast China, has long been impacted by climate change, with drought disasters significantly affecting its agricultural output and ecological environment. Accurate drought prediction is essential for the effective utilization of water resources and agricultural production. This study proposes a novel drought prediction model that utilizes the SSA-VMD (Sparrow Search Algorithm Optimized Variational Mode Decomposition) technique to decompose meteorological data, followed by the reconstruction of the decomposed components using four entropy algorithms, including approximate entropy. The model integrates ARIMA (AutoRegressive Integrated Moving Average) and BiLSTM (Bidirectional Long Short-Term Memory network) for forecasting the reconstructed components, subsequently combining their outputs. In terms of spatio-temporal data, the study employs prediction models including the spatio-temporal cube, spatio-temporal hotspot analysis integrated with empirical kriging, and local outlier analysis to examine spatial distribution. The model's predictive performance is validated from three perspectives: statistical characteristics of the indicators, comparison between predicted and observed values through prediction curve plots, and box plots. The results demonstrate that the combined SSA-VMD-ARIMA-BiLSTM model significantly enhances prediction accuracy compared to single models, as exemplified by its application in Changchun City. The model achieved an R2 of 0.938 and a root mean square error (RMSE) of 0.047 in drought prediction, outperforming the single ARIMA model (R2: 0.636, RMSE: 0.709) and the BiLSTM model (R2: 0.514, RMSE: 0.901). Additionally, across the entire province, the model's R2, MAE, and RMSE are 0.82, 0.15, and 0.083, respectively, suggesting that the model exhibits not only high prediction accuracy but also a degree of generalizability. Furthermore, the results from the spatio-temporal cube, spatio-temporal hotspot analysis, and local outlier analysis demonstrate the method's high accuracy and stability in predicting both short-term and long-term droughts. Particularly in short-term drought prediction, the model effectively captures the spatio-temporal distribution characteristics of short-term meteorological droughts. This study offers new methodological support for enhancing the early warning capabilities of drought risk in Jilin Province, providing a robust foundation for addressing the challenges posed by climate change. The findings not only address certain shortcomings in current drought prediction research but also introduce new methodologies and perspectives for future studies. [ABSTRACT FROM AUTHOR]

Monte Carlo rendering is widely used in the movie industry. Since it is costly to produce noise‐free results directly, Monte Carlo denoising is often applied as a post‐process. Recently, deep learning methods have been successfully leveraged in Monte Carlo denoising. They are able to produce high quality denoised results, even with very low sample rate, e.g. 4 spp (sample per pixel). However, for difficult scene configurations, some details could be blurred in the denoised results. In this paper, we aim at preserving more details from inputs rendered with low spp. We propose a novel denoising pipeline that handles three‐scale features ‐ pixel, sample and path ‐ to preserve sharp details, uses an improved Res2Net feature extractor to reduce the network parameters and a smooth feature attention mechanism to remove low‐frequency splotches. As a result, our method achieves higher denoising quality and preserves better details than the previous methods. [ABSTRACT FROM AUTHOR]

Gradient‐domain rendering can highly improve the convergence of light transport simulation using the smoothness in image space. These methods generate image gradients and solve an image reconstruction problem with rendered image and the gradient images. Recently, a previous work proposed a gradient‐domain volumetric photon density estimation for homogeneous participating media. However, the image reconstruction relies on traditional L1 reconstruction, which leads to obvious artifacts when only a few rendering passes are performed. Deep learning based reconstruction methods have been exploited for surface rendering, but they are not suitable for volume density estimation. In this paper, we propose an unsupervised neural network for image reconstruction of gradient‐domain volumetric photon density estimation, more specifically for volumetric photon mapping, using a variant of GradNet with an encoded shift connection and a separated auxiliary feature branch, which includes volume based auxiliary features such as transmittance and photon density. Our network smooths the images on global scale and preserves the high frequency details on a small scale. We demonstrate that our network produces a higher quality result, compared to previous work. Although we only considered volumetric photon mapping, it's straightforward to extend our method for other forms, like beam radiance estimation. [ABSTRACT FROM AUTHOR]

Inspection of a power transmission line by an unmanned aerial vehicle (UAV) is an important measure to ensure the operational safety of the power grid. A semantic segmentation model based on deep learning can assist UAVs in extracting power lines, which is a crucial task to ensure the safe navigation of UAVs. However, the existing models are vulnerable to complex background interference in aerial images and the thin structure of power lines. To solve these problems, an improved power line segmentation model based on Deeplabv3+ (PL-Deeplab) is proposed in this paper. To create our model, the multibranch concatenation network (MCNet) with stronger feature extraction capability is introduced to the backbone network. Then, we design a one-shot aggregation feature pyramid (OSAFP) to effectively extract the contextual information and provide a larger receptive field. Furthermore, the feature fusion module (FFM) in the decoder is used to increase the utilization of high-low level feature information and improve the ability of the model to accurately segment the boundaries of power lines. In addition, the weighted loss function is used to solve the problem of class imbalance in the power lines dataset. A large number of experiments on the power lines public dataset show that the proposed model can complete the power line inspection task better and provide security for the UAV power line inspection. [ABSTRACT FROM AUTHOR]