Deep learning based shadow removal aims to learn a mapping that translates shadow image from shadow domain to non-shadow domain. In this work, we develop a sim-ple yet effective target-consistency generative adversarial net-work (TC-ShadowGAN) for the task of unpaired shadow re-moval without any additional label for supervision. Com-pared with the popular bidirectional mapping based methods, TC-ShadowGAN targets to directly learn a one-side mapping to translate shadow images into shadow-free ones. With the proposed target-consistency constraint that is designed to connect a couple of GAN-based sub-networks, the corre-lations between shadow image and the output shadow-free image, and the realness of recovered shadow-free image are strictly confined. Extensive quantitative and qualitative evalu-ation experiments results show that TC-ShadowGAN outper-forms the state-of-the-art unpaired shadow removal methods by 14.9% in terms of FID and 31.5% in terms of KID on the unpaired shadowremoval dataset. The code isavailable at https://github.com/chaotan/TC-GAN.

With the highly demand of large-scale and real-time weather service for public, a refinement of short-time cloudage prediction has become an essential part of the weather forecast productions. To provide a weather-service-compliant cloudage nowcasting, in this paper, we propose a novel hierarchical Convolutional Long-Short-Term Memory network based deep learning model, which we term as FORECAST-CLSTM, with a new Forecaster loss function to predict the future satellite cloud images. The model is designed to fuse multi-scale features in the hierarchical network structure to predict the pixel value and the morphological movement of the cloudage simultaneously. We also collect about 40K infrared satellite nephograms and create a large-scale Satellite Cloudage Map Dataset(SCMD). The proposed FORECAST-CLSTM model is shown to achieve better prediction performance compared with the state-of-the-art ConvLSTM model and the proposed Forecaster Loss Function is also demonstrated to retain the uncertainty of the real atmosphere condition better than conventional loss function.

In this work, an iterative maximum entropy thresholding algorithm is proposed, due to the fact that the thresholds derived from the maximum entropy method are non-optimal. Firstly, the maximum entropy thresholding algorithm is applied to the grey images which are roughly grouped into background and object. Secondly, two Gaussian distributions are fitted according to the mean and variance of each group. Because the optimal threshold is located at their intersection of two Gaussian populations, the presented scheme optimizes thresholds using an iterative scheme until converging to the optimal position. Finally, extensive experiments are performed, and the results show that the thresholds derived from the maximum entropy scheme using the proposed iterative threshold optimization approach can be converged to the optimal position.

The emotion contagion is an important research point in the crowd simulation field. However, as the positions of crowd usually change dynamically, computing the contagion process becomes a challenge. There exists a solution for a similar problem, the N-body problem, but it's too time consuming to be adapted in crowd simulation as it needs to calculate the reactions between every two objects. In order to solve this problem, we provided a fast adjacent test and contagion computing algorithm with GPU acceleration. Firstly individuals' affection fields were set and projected onto two dimensional ground and represented by squares of the same size; secondly the whole ground was divided into squares as the same size as those for the affection projection; thirdly, adjacent squares were tested to generate the nearer neighbors and the contagion results were calculated. All these steps were processed in parallel by GPU.