A Symmetric Fusion Learning Model for Detecting Visual Relations and Scene Parsing

Table 3

Ablation studies on the key components of our method. We report the phrase and relationship detection performance in R@n scores (R@50 and R@100).

K	Method	Relation detection		Phrase detection
		R@50	R@100	R@50	R@100
1	Baseline	25.39	29.67	31.50	37.00
	Baseline +	25.62	29.92	31.45	37.21
	Baseline +  +	26.01	29.90	32.02	37.31
70	Baseline	27.38	34.33	33.56	41.90
	Baseline +	27.38	34.81	33.91	42.46
	Baseline +  +	28.63	35.21	34.88	43.07