Generative adversarial networks (GANs) have been widely applied in the field of meteorological research, particularly in the downscaling of images due to their ability to generate super-resolution images. In recent years, numerous researchers have combined GANs with recurrent neural networks (RNNs) to address the issue of meteorological super-resolution. However, these models do not take into account the spatial variations of meteorological sequences. In this paper, we propose a super-resolution method named RN-STLSTM-GAN, which combines GANs with RN-STLSTM and ESA networks to learn the spatiotemporal features of meteorological sequences. Specifically, we first apply the RN-STLSTM at the initialization of the generator and discriminator to learn the spatiotemporal relationships between sequential images. Second, an ESA network is combined with the RN-STLSTM structure to enhance the learning of spatial features. Thirdly, LeakyReLU is used as the activation function for both the generator and discriminator to minimize the loss of image data during model training. Experiments conducted on the NJU-CPOL datasets demonstrate that our proposed method outperforms other existing methods and can generate realistic and temporally consistent super-resolution sequences for datasets at different heights.

This paper addresses the problem of recognizing human actions captured with depth cameras. Human action recognition is a challenging task as the articulated action data is high dimensional in both spatial and temporal domains. An effective approach to handle this complexity is to divide human body into different body parts according to human skeletal joint positions, and performs recognition based on these part-based feature descriptors. Since different types of features could share some similar hidden structures, and different actions may be well characterized by properties common to all features (sharable structure) and those specific to a feature (specific structure), we propose a joint group sparse regression-based learning method to model each action. Our method can mine the sharable and specific structures among its part-based multiple features meanwhile imposing the importance of these part-based feature structures by joint group sparse regularization, in favor of discriminative part-based feature structure selection. To represent the dynamics and appearance of the human body parts, we employ part-based multiple features extracted from skeleton and depth data respectively. Then, using the group sparse regularization techniques, we have derived an algorithm for mining the key part-based features in the proposed learning framework. The resulting features derived from the learnt weight matrices are more discriminative for multi-task classification. Through extensive experiments on three public datasets, we demonstrate that our approach outperforms existing methods.

This paper addresses the problem of recognizing human skeletal interactions using multiview data captured from depth sensors. The interactions among people are important cues for group and crowd human behavior analysis. In this paper, we focus on modeling the person-person skeletal interactions for human activity recognition. First, we propose a novel graph model in each single-view case to encode class-specific person-person interaction patterns. Particularly, we model each person-person interaction by an attributed graph, which is designed to preserve the complex spatial structure among skeletal joints according to their activity levels as well as the spatio-temporal joint features. Then, combining the graph models for each single-view case, we propose the multigraph model to characterize each multiview interaction. Finally, we apply a general multiple kernel learning method to determine the optimal kernel weights for the proposed multigraph model while the optimal classifier is jointly learned. We evaluate the proposed approach on the M2I dataset, the SBU Kinect interaction dataset, and our interaction dataset. The experimental results show that our proposed approach outperforms several existing interaction recognition methods.