On the SUIM dataset (Semantic Segmentation of Underwater Imagery: Dataset and Benchmark), the performance of UNet (IoU:56.9700) is significantly lower than that of FCN(66.75) !!! In fact, the performance of FCN is consistent with that mentioned in the paper, but the performance of UNet differs greatly. Can you provide the same network as the original UNet paper?
On the SUIM dataset (Semantic Segmentation of Underwater Imagery: Dataset and Benchmark), the performance of UNet (IoU:56.9700) is significantly lower than that of FCN(66.75) !!! In fact, the performance of FCN is consistent with that mentioned in the paper, but the performance of UNet differs greatly. Can you provide the same network as the original UNet paper?