Table 4: Comparison of bowel inflammation detection results between the baseline and the proposed methods for Label 3 GT.

Label 3 GT	mAP (%)		Precision (%)		Recall (%)
Methods	@10	@50	@10	@50	@10	@50
SSSL (Baseline)	68.15	62.49	65.15	58.29	69.85	65.45
MSSL (s = 1)	62.06	55.82	58.15	50.9	64.05	59.24
MSSL (s = 2)	62.67	57.28	59.13	53.3	64.6	60.13
MSSL (s = 3)	63.22	59.41	60.66	56.39	64.79	61.65
MSML (s = 1)	71.01	66.21	69.45	63.29	72.36	68.24
MSML (s = 2)	65.45	61.08	63.13	57.89	67.06	63.39
MSML (s = 3)	74.18	71.38	73.86	70.25	75.07	72.59