Table 8.

Class-wise performance (F1-score) on coarse-grained medical instructional videos of MedVidCL dataset using the best monomodel and multimodal models from Table 6.

Models		Pain	Oral Health	Brain & Nerves	Infection	Equipment	ENT	Musculoskeletal Health	First Aid	Eye	Cardiovascular Health	Surgery	Respiratory	Systemic
(1)	BigBird-Base	35.90	85.11	45.16	00.00	44.00	72.97	73.09	64.48	70.18	26.23	00.00	70.59	33.33
(2)	ViT + Transformer	21.79	57.83	4.55	00.00	45.10	22.41	66.25	50.91	14.29	11.63	11.76	26.09	3.51
(3)	L + V (ViT) + Transformer	21.47	55.91	6.45	00.00	45.74	22.07	63.09	50.00	4.35	10.08	21.05	28.57	2.63