Monocular 3D face reconstruction plays a crucial role in avatar generation, with significant demand in web-related applications such as generating virtual financial advisors in FinTech. Current reconstruction methods predominantly rely on deep learning techniques and employ 2D self-supervision as a means to guide model learning. However, these methods encounter challenges in capturing the comprehensive 3D structural information of the face due to the utilization of 2D images for model training purposes. To overcome this limitation and enhance the reconstruction of 3D structural features, we propose an innovative approach that integrates existing 2D features with 3D features to guide the model learning process. Specifically, we introduce the 3D-ID Loss, which leverages the high-dimensional structure features extracted from a Spectral-Based Graph Convolution Encoder applied to the facial mesh. This approach surpasses the sole reliance on the 3D information provided by the facial mesh vertices coordinates. Our model is trained using 2D-3D data pairs from a combination of datasets and achieves state-of-the-art performance on the NoW benchmark.
The 24th Winter Olympics was held in Beijing, and the air quality in the Beijing area has become the focus of the world’s attention. The Beijing government has taken a series of strict measures to control pollutant emissions during the Winter Olympics, which also provides us with a valuable opportunity to study the impact of meteorological conditions on pollutants. We defined November, December, January, February, and March as the polluted period in Beijing, and used the T-PCA method to divide the circulation types (CTs) affecting Beijing into six kinds (CT1-CT6). It was found that under the control of the western high pressure (CT1) and the northwest high pressure (CT4), the concentrations of PM 2.5 , NO 2 , SO 2 and CO in Beijing were lower; while under the control of the northern high pressure (CT2), eastern high pressure (CT5), southeast high (CT3) and northeast low pressure (CT6), the concentration of PM 2.5 , NO 2 , SO 2 and CO is higher. By analyzing the daily CTs, wind field and pollutant concentration changes in the Beijing area during the Beijing Winter Olympics, it was found that when two pollution events occurred during the Winter Olympics, the Beijing area was mainly prevailed by CT2, CT3, and CT6. Comparing the frequency of occurrence of six CTs during the 2022 Winter Olympics and the same period from 2014 to 2021, it was found that the proportion of CT1 and CT4 increased significantly during the Winter Olympics. Finally, the FLEXPART-WRF model was used to analyze the 48-h backward footprint distribution of pollutant particles in Beijing during the Winter Olympics. It further showed that the circulation in the Beijing area during the Winter Olympics was generally conducive to the dispersion of pollutants, and the air quality was better.
Energy Internet (EI) has drawn increasingly interests in related industries. The implement of EI is a milestone on the road-map to promote the revolution of both energy production and consumption, which is important to achieve sustainable energy development. This paper investigates the possibility on establishing the EI by integrating application-level intelligent management system with traditional energy grid in neighbourhood scale. As a new energy utilisation model, EI realises the deep integration between energy industry and information and communications technology (ICT) in order to optimise regional energy management and optimal dispatch. With the support of Internet of things technology and big data technology, this system plays an important role in coordinating management and scheduling while realising the interaction between market entities and end-users at the same time. This paper reviews and conducts in-depth research on the background, concept and technical model of EI, and also discusses the implementation of EI in a practical industrial parks.
Monocular 3D face reconstruction plays a crucial role in avatar generation, with significant demand in web-related applications such as generating virtual financial advisors in FinTech. Current reconstruction methods predominantly rely on deep learning techniques and employ 2D self-supervision as a means to guide model learning. However, these methods encounter challenges in capturing the comprehensive 3D structural information of the face due to the utilization of 2D images for model training purposes. To overcome this limitation and enhance the reconstruction of 3D structural features, we propose an innovative approach that integrates existing 2D features with 3D features to guide the model learning process. Specifically, we introduce the 3D-ID Loss, which leverages the high-dimensional structure features extracted from a Spectral-Based Graph Convolution Encoder applied to the facial mesh. This approach surpasses the sole reliance on the 3D information provided by the facial mesh vertices coordinates. Our model is trained using 2D-3D data pairs from a combination of datasets and achieves state-of-the-art performance on the NoW benchmark.
Severe haze events have many adverse effects on agricultural production and human activity. Haze events are often associated with specific patterns of atmospheric circulation. Therefore, studying the relationship between atmospheric circulation and haze is particularly important for early warning and forecasting of urban haze events. In order to study the relationship between multi-scale atmospheric circulation and severe haze events in autumn and winter in Shanghai, China, we used a T-mode objective classification method to classify autumn and winter atmospheric circulation patterns into six types based on sea level pressure data from 2007 to 2016 in the Shanghai area. For the period between September 2016 and February 2017, we used the Allwine–Whiteman method to classify the local wind in Shanghai into three categories: stagnation, recirculation, and ventilation. By further studying the PM2.5 concentration distribution, visibility distribution, and other meteorological characteristics of each circulation type (CT) and local wind field type, we found that the Shanghai area is most prone to severe haze when exposed to certain circulation patterns (CT1, CT2, and CT4), mainly associated to the cold air activity and the displacement of the high pressure system relative to Shanghai. We also found that the local wind fields in the Shanghai area are dominated by recirculation and stagnation events. These conclusions were further verified by studying a typical pollution process in Shanghai in November 2016 and the pollutant diffusion path using the HYSPLIT (Hybrid Single Particle Lagrangian Integrated Trajectory model) simulation model.
During the lockdown implemented to curb the spread of COVID-19, human activities have drastically reduced, providing a valuable opportunity to study and compare the impact of meteorological conditions and human activities on air quality. In this study, large-scale weather circulation, local meteorological conditions, and the impact of human activities are comprehensively considered, and changes in the concentration of major air pollutants in the northeast during this period are systematically studied. The large-scale weather circulation patterns that mainly affect the northeast region are divided into nine types by using the T-mode Principal components analysis objective circulation classification method. It is found that the northeast region is located at the edge of weak high pressure (Types 1, 2, and 7) and at the rear of high pressure (Type 4) and has higher concentrations of PM2.5, NO 2 , SO 2 , and CO; in cyclonic weather systems, low vortices (Types 3 and 5) and under the influence of the updraft (Type 6) in front of the trough, the ozone concentration is higher. The changes in the concentrations of PM2.5, NO 2 , CO, SO 2 , and O 3 in the three cities, namely Shenyang, Changchun, and Harbin, during the lockdown period are compared, and it is found that the concentrations of PM2.5, NO 2 , CO, and SO 2 have a tendency to first decrease and then increase, while the changes of O 3 concentration are cyclical and increased significantly during this period. This demonstrates that pollutants such as PM2.5, NO 2 , CO, and SO 2 are more susceptible to human activities and local meteorological conditions, and changes in O 3 concentration are more closely related to changes in weather circulation types. Finally, the FLEXPART-WRF model is used to simulate the pollution process of nine circulation types, which confirms that particulate pollution in the northeast is mainly affected by local emissions and local westward sinking airflow.
From the perspective of energy providers, accurate short-term load forecasting plays a significant role in the energy generation plan, efficient energy distribution process and electricity price strategy optimisation. However, it is hard to achieve a satisfactory result because the historical data is irregular, non-smooth, non-linear and noisy. To handle these challenges, in this work, we introduce a novel model based on the Transformer network to provide an accurate day-ahead load forecasting service. Our model contains a similar day selection approach involving the LightGBM and k-means algorithms. Compared to the traditional RNN-based model, our proposed model can avoid falling into the local minimum and outperforming the global search. To evaluate the performance of our proposed model, we set up a series of simulation experiments based on the energy consumption data in Australia. The performance of our model has an average MAPE (mean absolute percentage error) of 1.13, where RNN is 4.18, and LSTM is 1.93.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
'/%3e%3c/defs%3e%3cpath fill='%23012169' d='M0 0h10080v5040H0z'/%3e%3cpath stroke='%23fff' stroke-width='.6' d='m0 0 6 3m0-3L0 3' clip-path='url(%23b)' transform='scale(840)'/%3e%3cpath stroke='%23e4002b' stroke-width='.4' d='m0 0 6 3m0-3L0 3' clip-path='url(%23c)' transform='scale(840)'/%3e%3cpath stroke='%23fff' stroke-width='840' d='M2520 0v2520M0 1260h5040'/%3e%3cpath stroke='%23e4002b' stroke-width='504' d='M2520 0v2520M0 1260h5040'/%3e%3cg fill='%23fff'%3e%3cuse xlink:href='%23d' x='2520' y='3780'/%3e%3cuse xlink:href='%23a' x='7560' y='4200'/%3e%3cuse xlink:href='%23a' x='6300' y='2205'/%3e%3cuse xlink:href='%23a' x='7560' y='840'/%3e%3cuse xlink:href='%23a' x='8680' y='1869'/%3e%3cuse xlink:href='%23e' x='8064' y='2730'/%3e%3c/g%3e%3c/svg%3e)
'/%3e%3cuse xlink:href='%23a' transform='translate(288.889 -214.211)'/%3e%3cuse xlink:href='%23a' transform='translate(108 342.749)'/%3e%3cuse xlink:href='%23a' transform='translate(469.189 342.749)'/%3e%3c/svg%3e)