Learning comprehensive spatiotemporal features is crucial for human action recognition. Existing methods tend to model the spatiotemporal feature blocks in an integrate-separate-integrate form, such as appearance-and-...Learning comprehensive spatiotemporal features is crucial for human action recognition. Existing methods tend to model the spatiotemporal feature blocks in an integrate-separate-integrate form, such as appearance-and-relation network(ARTNet) and spatiotemporal and motion network(STM). However, with blocks stacking up, the rear part of the network has poor interpretability. To avoid this problem, we propose a novel architecture called spatial temporal relation network(STRNet), which can learn explicit information of appearance, motion and especially the temporal relation information. Specifically, our STRNet is constructed by three branches,which separates the features into 1) appearance pathway, to obtain spatial semantics, 2) motion pathway, to reinforce the spatiotemporal feature representation, and 3) relation pathway, to focus on capturing temporal relation details of successive frames and to explore long-term representation dependency. In addition, our STRNet does not just simply merge the multi-branch information, but we apply a flexible and effective strategy to fuse the complementary information from multiple pathways. We evaluate our network on four major action recognition benchmarks: Kinetics-400, UCF-101, HMDB-51, and Something-Something v1, demonstrating that the performance of our STRNet achieves the state-of-the-art result on the UCF-101 and HMDB-51 datasets, as well as a comparable accuracy with the state-of-the-art method on Something-Something v1 and Kinetics-400.展开更多
Accurate prediction of solar photovoltaic power is crucial for renewable integration.However,existing methods struggles to simultaneously capture its complex spatial correlation and nonlinear temporal dependencies.Her...Accurate prediction of solar photovoltaic power is crucial for renewable integration.However,existing methods struggles to simultaneously capture its complex spatial correlation and nonlinear temporal dependencies.Here,we propose a novel spatiotemporal relationship fusion network(STRFN)for short-term prediction of photovoltaic power generation.STRFN uses convolutional neural networks to extract spatial features,long short-term memory networks to capture time dependence,and an attention mechanism to enhance its expressiveness.Additionally,the optimal network hyperparameters of STRFN are identified through Bayesian optimization.Moreover,it employs advanced data preprocessing techniques to improve input data quality.These techniques include feature recognition,principal component analysis,location coding,and sliding window segmentation.Our STRFN is applied to two typical PV systems for demonstration and compared with traditional deep learning models.The results show that our model’s accuracy and stability significantly outperform traditional deep learning models,with RMSE of 2.46 and 0.036,and MAPE of 1.51%and 1.94%.Furthermore,in predictions for typical days across four seasons,our STRFN still maintained consistent superior performance-evidenced by its normalized RMSE(NRMSE)of 0.90%and 0.61%for the two PV systems.Finally,we integrate data processing,model training,and results visualization together into a one-stop platform and make it user-friendly and easily improved for solar power prediction.Our proposed method along with its forecasting platform can offer valuable insights and guidelines for researchers and PV operators.展开更多
基金supported by National Natural Science Foundation of China(Nos.U1836218,62020106012,61672265 and 61902153)the 111 Project of Ministry of Education of China(No.B12018)+1 种基金the EPSRC Programme FACER2VM(No.EP/N007743/1)the EPSRC/MURI/Dstl Project under(No.EP/R013616/1.)。
文摘Learning comprehensive spatiotemporal features is crucial for human action recognition. Existing methods tend to model the spatiotemporal feature blocks in an integrate-separate-integrate form, such as appearance-and-relation network(ARTNet) and spatiotemporal and motion network(STM). However, with blocks stacking up, the rear part of the network has poor interpretability. To avoid this problem, we propose a novel architecture called spatial temporal relation network(STRNet), which can learn explicit information of appearance, motion and especially the temporal relation information. Specifically, our STRNet is constructed by three branches,which separates the features into 1) appearance pathway, to obtain spatial semantics, 2) motion pathway, to reinforce the spatiotemporal feature representation, and 3) relation pathway, to focus on capturing temporal relation details of successive frames and to explore long-term representation dependency. In addition, our STRNet does not just simply merge the multi-branch information, but we apply a flexible and effective strategy to fuse the complementary information from multiple pathways. We evaluate our network on four major action recognition benchmarks: Kinetics-400, UCF-101, HMDB-51, and Something-Something v1, demonstrating that the performance of our STRNet achieves the state-of-the-art result on the UCF-101 and HMDB-51 datasets, as well as a comparable accuracy with the state-of-the-art method on Something-Something v1 and Kinetics-400.
基金National Key Research and Development Program of China(2024YFB4006400).
文摘Accurate prediction of solar photovoltaic power is crucial for renewable integration.However,existing methods struggles to simultaneously capture its complex spatial correlation and nonlinear temporal dependencies.Here,we propose a novel spatiotemporal relationship fusion network(STRFN)for short-term prediction of photovoltaic power generation.STRFN uses convolutional neural networks to extract spatial features,long short-term memory networks to capture time dependence,and an attention mechanism to enhance its expressiveness.Additionally,the optimal network hyperparameters of STRFN are identified through Bayesian optimization.Moreover,it employs advanced data preprocessing techniques to improve input data quality.These techniques include feature recognition,principal component analysis,location coding,and sliding window segmentation.Our STRFN is applied to two typical PV systems for demonstration and compared with traditional deep learning models.The results show that our model’s accuracy and stability significantly outperform traditional deep learning models,with RMSE of 2.46 and 0.036,and MAPE of 1.51%and 1.94%.Furthermore,in predictions for typical days across four seasons,our STRFN still maintained consistent superior performance-evidenced by its normalized RMSE(NRMSE)of 0.90%and 0.61%for the two PV systems.Finally,we integrate data processing,model training,and results visualization together into a one-stop platform and make it user-friendly and easily improved for solar power prediction.Our proposed method along with its forecasting platform can offer valuable insights and guidelines for researchers and PV operators.
