Scaling Human-Object Interaction Recognition in the Video through Zero-Shot Learning

Table 2

The effect of local feature aggregation on HOI recognition performance. The model was trained with 1A + 2B and tested on 2A + 1B and all data.

Method	mAP (%)
	ALL data	Unseen data (2A + 1B)
2Stream – WE - VLAD	17.8	14.83
2Stream + WE - VLAD	19.5	16.08
2Stream – WE + VLAD (rnn)	18.7	15.33
2Stream + WE + VLAD (rnn)	20.96	16.96
2Stream + WE + VLAD (cnn)	20.65	16.65