Automatic Traffic State Recognition Based on Video Features Extracted by an Autoencoder

Table 2

Structural parameters of the eight candidate autoencoder models.

Model number	Input data dimension	Number of autoencoder hidden layers	Dimension of the dimension-reduced data	Autoencoder structure
I	32 × 32	5	3	(32 × 32,256), (256,128), (128,64), (64,12), (12,3)
II	32 × 32	5	4	(32 × 32,256), (256,128), (128,64), (64,12), (12,4)
III	32 × 32	5	5	(32 × 32,256), (256,128), (128,64), (64,12), (12,5)
IV	32 × 32	5	6	(32 × 32,256), (256,128), (128,64), (64,12), (12,6)
V	64 × 64	5	3	(64 × 64,256), (256,128), (128,64), (64,12), (12,3)
VI	64 × 64	5	4	(64 × 64,256), (256,128), (128,64), (64,12), (12,4)
VII	64 × 64	5	5	(64 × 64,256), (256,128), (128,64), (64,12), (12,5)
VIII	64 × 64	5	6	(64 × 64,256), (256,128), (128,64), (64,12), (12,6)