Depth neural network (DNN) has become a research hotspot in the field of image recognition. Developing a suitable solution to introduce effective operations and layers into DNN model is of great significance to improve the performance of image and video recognition. To achieve this, through making full use of block information of different sizes and scales in the image, a multiscale pooling deep convolution neural network model is designed in this paper. No matter how large the feature map is, multiscale sampling layer will output three fixed-size character matrices. Experimental results demonstrate that this method greatly improves the performance of the current single training image, which is suitable for solving the image generation, style migration, image editing, and other issues. It provides an effective solution for further industrial practice in the fields of medical image, remote sensing, and satellite imaging.

中文翻译：

---

基于多尺度池深度卷积神经网络的图像识别

深度神经网络（DNN）已成为图像识别领域的研究热点。开发合适的解决方案以将有效的操作和层次引入DNN模型中对于提高图像和视频识别的性能具有重要意义。为此，通过充分利用图像中不同大小和尺度的块信息，设计了一种多尺度池深度卷积神经网络模型。无论特征图有多大，多尺度采样层都将输出三个固定大小的字符矩阵。实验结果表明，该方法大大提高了当前单训练图像的性能，适用于解决图像生成，样式迁移，图像编辑等问题。