error correctionFigure 6 shows the visualization results of each method based on dataset #2 We can see that a small part of each category in the segmentation results of Swin-Unet and TransuUNet are mi

Figure 6 shows the visualization results of each method based on dataset #2. We can see that a small part of each category in the segmentation results of Swin-Unet and TransuUNet is misjudged, resulting in the segmentation results not corresponding to the ground truth. Additionally, there are many edge burrs in the segmentation results of Swin-Unet. Similarly, there are many misjudgments in the segmentation results of U-Net, PSPnet, and DeepLabv3+, leading to missing and incorrect segmentation results for many categories. In contrast, our proposed segmentation method, RTMOSeg, is very close to the ground truth, and the segmentation details are excellent