为节省时间和研究人员的精力，采用计算机辅助对细胞或细胞核进行检测。利用卷积神经网络的U-net衍生网络并结合图像处理过程开展对细胞核的检测。研究结果表明，该检测方法的检测精确度为0.82，召回率为0.83，F指标为0.83，具有较好的识别和分割效果。

In order to save time and researchers' energy, computers are used to assist in the detection of cells or nuclei. In this paper, a derivative network U-net of convolution neural network is combined with image processing to detect nucleus. The result shows that the accuracy is 0.82, the recall rate is 0.83, and the F index is 0.83, which has good recognition and segmentation effect.