We analyzed the network structure of real-time object detection models and found that features in feature concatenation stage are very rich. Applying an attention module here can effectively improve accuracy model. However, commonly used or self-attention shows poor performance inference efficiency. Therefore, we propose a novel module, called 2D local superimposed self-attention, for neck netw...

Abstract In our study we assess the responsiveness of Hungarian local governments to requests for information by Roma and non-Roma clients, relying on a nationwide correspondence study. Our paper has both methodological substantive relevance. The novelty is that treat discrimination as classification problem what extent emails written clients can be distinguished, which in turn serves metric ge...

Remote sensing image scene classification is an important task of remote interpretation, which has recently been well addressed by the convolutional neural network owing to its powerful learning ability. However, due multiple types geographical information and redundant background images, most CNN-based methods, especially those based on a single CNN model ignoring combination global local feat...