Vision Transformer-Based Video Hashing Retrieval for Tracing the Source of Fake Videos

Table 5

Evaluation of classification and localization on different types of datasets.

Datasets	Localization (mIoU)			Classification (%)
	DMAC [62]	DMVN [63]	Ours	ACC	Total originals
DFD	—	—	0.726	94.9	363
DFS	—	—	0.880	100.0	133
Video splicing	0.828	0.751	0.842	100.0	30
DAVIS-VI	0.828	0.751	0.882	100.0	50

Bold values represent the best results.