Table 3. Quantitative results on the public dataset Kvasir-instrument (numbers in bold are the best indicator results).

Method	mIoU (%)	Mean F1 (%)	Mean precision (%)	Mean recall (%)
UNet	91.91	95.66	95.93	95.40
TransUNet	91.60	95.48	96.08	94.90
CGNet	92.59	96.05	96.76	95.38
SegFormer	91.57	95.46	96.08	94.87
GCNet	92.14	95.79	96.67	94.96
ContextNet	91.84	95.62	96.44	94.84
DABNet	93.27	96.43	96.91	95.95
FPENet	92.05	95.75	96.36	95.15
Ours	93.29	96.45	97.01	95.90