Correction: AWUNet: leaf area segmentation based on attention gate and wavelet pooling mechanism

Accurate segmentation and detection of leaf area is of great significance for automatic plant growth monitoring and is one of the most effective strategies to monitor the sustainability of plant growth, yield, width, and height of plants for agricultural production. The leaf segmentation in carrot plants is a challenging task because of secondary leaflets, which introduces complex variability in leaf shape. In this paper, a novel AWUNet (Attention-gated Wavelet pooled UNet) integrating the concepts of wavelet pooling to reduce the size of the feature map and attention gate module which focuses the semantic content in the feature map is proposed. The skip connections in existing UNet (U-shaped network) model are remodeled using attention gate mechanism for saliency improvement, and pooling layers utilize wavelet function for compression. The effectiveness of hyper parameters is investigated to enhance the proposed AWUNet model’s accuracy. The proposed model was evaluated and compared with existing methods such as UNet + + , Inception UNet, SAUNet, and INCSA UNet, as well as pre-trained deep learning architectures like Visual Geometry Group models such as VGG16-UNet, VGG19-UNet, and residual neural network (ResNet) UNet. The highest segmentation IoU (intersection of union) score of 94.81% was achieved by the proposed AWUNet model, whereas the state-of-the-art methods UNet, UNet + + , Inception UNet, SAUNet, INCSA UNet,VGG16 UNet,VGG19 UNet, and ResNet UNet achieved IoU score of 86.29%, 63.37%, 45.26%, 72.39%, 89.30%, 82.99%, 85.16%, and 75.50%, respectively. The findings showed that the proposed AWUNet model outperforms than other models in segmenting leaf area in plants and can be implemented in the agriculture sector for crop monitoring.