Vessel Segmentation on Fundus-AVSeg combined vasculature (test)
0.8352IoUU-Net++ w/ Attn
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| U-Net++ w/ AttnLoss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.8352 | 0.91 | 3.59 | 0.9405 | 0.8831 | |
| U-Net++Loss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.8169 | 0.8989 | 4.27 | 0.9519 | 0.8536 | |
| Attention U-NetLoss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.8145 | 0.8975 | 4.6 | 0.934 | 0.8655 | |
| SegResNetLoss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.7908 | 0.8829 | 7.05 | 0.9405 | 0.8336 | |
| SwinUNETRLoss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.7476 | 0.8554 | 9.97 | 0.9504 | 0.7794 | |
| TransUNet (ViT-B)Loss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.7329 | 0.8456 | 8.34 | 0.9019 | 0.7975 | |
| SegformerLoss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.66 | 0.7948 | 7.78 | 0.8519 | 0.7472 | |
| U-NetLoss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.6286 | 0.771 | 12.45 | 0.9411 | 0.6567 | |
| DeepLabV3+Loss=BCE loss, Optimizer=AdamW, Augmentation=None, Pre-trained encoder=None2026.01 | 0.4902 | 0.6576 | 11 | 0.7328 | 0.5977 |