U-Net Pyramid CBAM Gate (UPCG) deplys both CBAM and attention gate in sequence. Experiments show more attention can lead to better accuracy, especially in difficult scenarios( e.g. in the borders ). UPCG managed to beat successful U-Nets such as U-Net-ResNet50, Double U-Net, and R2U-Net, which are also investigated in this study. Also, 4 different augmentation configurations consisting of our hair removal, our hair augmentation, and cutmix and mosaic augmentation along with geometrical augmentations offered by Albumentations, and 10 different U-Net-based architectures, with or without attention modules, including our UPCG were studied.