Visual Experience-Based Question Answering with Complex Multimodal Environments

Table 4

Performance analysis of scene graph generation depending on different state recognition models.

Models	Object mAP (%)	SGGen (%)
		Attribute	Relation	Total
	68.79	56.87	46.79	53.73
	85.12	69.69	60.09	67.35
	85.12	84.61	60.09	79.62
	100.0	100.0	95.89	98.80