Zongwei Zhou, Md Mahfuzur Rahman Siddiquee, Nima Tajbakhsh, Jianming Liang
Implementation of different kinds of Unet Models for Image Segmentation - Unet , RCNN-Unet, Attention Unet, RCNN-Attention Unet, Nested Unet
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Medical Image Segmentation | Kvasir-SEG | Average MAE | 0.048 | U-Net++ |
| Medical Image Segmentation | Kvasir-SEG | S-Measure | 0.862 | U-Net++ |
| Medical Image Segmentation | Kvasir-SEG | max E-Measure | 0.91 | U-Net++ |
| Medical Image Segmentation | Kvasir-SEG | mean Dice | 0.821 | U-Net++ |
| Medical Image Segmentation | 2018 Data Science Bowl | Dice | 0.8974 | Unet++ |
| Medical Image Segmentation | 2018 Data Science Bowl | mIoU | 0.9255 | Unet++ |
| Medical Image Segmentation | CVC-ClinicDB | mean Dice | 0.794 | U-Net++ |
| Medical Image Segmentation | SUN-SEG-Easy (Unseen) | Sensitivity | 0.457 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Hard | Dice | 0.554 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Hard | S-Measure | 0.685 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Hard | mean E-measure | 0.697 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Hard | mean F-measure | 0.544 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Hard | weighted F-measure | 0.48 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Hard (Unseen) | Sensitivity | 0.467 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Easy | Dice | 0.559 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Easy | S measure | 0.684 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Easy | mean E-measure | 0.687 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Easy | mean F-measure | 0.553 | UNet++ |
| Medical Image Segmentation | SUN-SEG-Easy | weighted F-measure | 0.491 | UNet++ |
| Semantic Segmentation | Cityscapes val | mIoU | 75.5 | UNet++ (ResNet-101) |
| Semantic Segmentation | AI-TOD | Dice | 70.19 | Unet++(ResNet-50) |
| Object Detection | PCOD_1200 | S-Measure | 0.801 | UNet++ |
| 3D | PCOD_1200 | S-Measure | 0.801 | UNet++ |
| Camouflaged Object Segmentation | PCOD_1200 | S-Measure | 0.801 | UNet++ |
| Object Segmentation | PCOD_1200 | S-Measure | 0.801 | UNet++ |
| 2D Classification | PCOD_1200 | S-Measure | 0.801 | UNet++ |
| 2D Object Detection | PCOD_1200 | S-Measure | 0.801 | UNet++ |
| 10-shot image generation | Cityscapes val | mIoU | 75.5 | UNet++ (ResNet-101) |
| 10-shot image generation | AI-TOD | Dice | 70.19 | Unet++(ResNet-50) |
| 16k | PCOD_1200 | S-Measure | 0.801 | UNet++ |