Table 1.

Darknet-53 network structure.

	Type	Number of Filters	Size	Stride	Output
	Convolutional	32	3 × 3	1	256 × 256
	Convolutional	64	3 × 3	2	128 × 128
1×	Convolutional	32	1 × 1	1
	Convolutional	64	3 × 3	1
	Residual	---	---	---	128 × 128
	Convolutional	128	3 × 3	2	64 × 64
2×	Convolutional	64	1 × 1	1
	Convolutional	128	3 × 3	1
	Residual	---	---	---	64 × 64
	Convolutional	256	3 × 3	2	32 × 32
8×	Convolutional	128	1 × 1	1
	Convolutional	256	3 × 3	1
	Residual	---	---	---	32 × 32
	Convolutional	512	3 × 3	2	16 × 16
8×	Convolutional	256	1 × 1	1
	Convolutional	512	3 × 3	1
	Residual	---	---	---	16 × 16
	Convolutional	1024	3 × 3	2	8 × 8
4×	Convolutional	512	1 × 1	1
	Convolutional	1024	3 × 3	1
	Residual	---	---	---	8 × 8
	Avgpool Connected Softmax	Global 1000