Title: Image segmentation based on improved UNET++
Abstract: In order to help researchers to perform cell segmentation, computer vision techniques are used to perform cell segmentation on neural cells. The addition of the Attention mechanism based on U-net++ suppresses the activity of irrelevant regions and improves the segmentation efficiency and segmentation accuracy. According to the experimental data, the segmentation accuracy of the method reaches 0.78 and the loss rate is 0.3, which can achieve a better segmentation effect compared with the traditional algorithm.