Table 7.

Surgical phase recognition few-shot results (mean accuracy)

	Train Set	ResNet50	ViT ImageNet	EndoViT
Stage 1	2 Videos	47.51 ± 1.33	63.59 ± 1.07	67.04 ± 2.92
	4 Videos	57.80 ± 2.67	67.72 ± 0.90	71.80 ± 0.49
	8 Videos	63.71 ± 1.48	75.50 ± 0.32	75.30 ± 1.83
Stage 2	2 Videos	68.23 ± 1.10	77.05 ± 1.71	78.89 ± 1.26
	4 Videos	74.50 ± 1.76	80.00 ± 0.62	80.28 ± 0.71
	8 Videos	77.43 ± 1.68	84.10 ± 0.38	84.68 ± 1.25