Research on Hyperspectral Image Target Detection By the Convergence Neural Network Based on Transfer Learning

This paper mainly proposes a hyperspectral image object detection algorithm based on transfer learning for Convolutional Neural Networks(CNN). Hyperspectral image object detection is one of the research hotspots in the field of image processing. With the rise of deep learning, more and more scholars have begun to study the application of deep learning in the field of hyperspectral object detection. However, it costs a lot of time for the models based on deep learning to train networks and adjust parameters. This paper mainly studies the correlation between different data sets. It hopes to find the mapping relationship between different data sets by using transfer learning, so as to avoid the time cost of training network. Finally, the paper tests in the PaviaU and PaviaC datasets, and proves that the transfer algorithm proposed in this paper can make the dataset achieve good detection results after transfer.