中图分类号： TP391.4
文献标识码： A
DOI： 10.19358/j.issn.2096-5133.2020.11.009
引用格式： 万川，王正勇，何海波，等. 基于改进P-Unet模型的岩屑颗粒识别[J].信息技术与网络安全，2020，39(11)：56-61.

Identification of cuttings particle based on improved P-Unet model

Abstract： This paper proposes a model based on improved P-Unet cuttings particle identification method. The method is based on the Unet model structure，and uses pyramid pooling module to aggregate contextual feature information in different regions to make full use of global information. The improved P-Unet model uses the residual network ResNeXt101，which reduces the number of super parameters on the premise of improving the accuracy of debris particle recognition. In this model，the focus loss function is used to solve the problem of the unbalance of cuttings particles. At the same time，the deep separable convolution is used to replace the traditional convolution，which greatly reduces the parameters of the network and the prediction time. The experimental results show that the recognition accuracy of the improved P-Unet model is improved compared with similar advanced algorithms，and the recognition results of cuttings particles are more accurate.

Key words : identification of cuttings particles；Unet network；P-Unet network；pyramid pooling module；deep separable convolution