YoloV2 darknet网络参数计算

YoloV2 darknet网络参数计算

Name	Filters	Output Dimension	Params	Flops
Conv1	3x3x3，32，strides=2，padding=1	224x224x32	0.896K	0.0867B
Max pool1	2x2，strides=2	112x112x32
Conv2	3x3x32，64，strides=1，padding=1	112x112x64	18.496k	0.4624B
Max pool2	2x2，strides=2	56x56x64
Conv3	3x3x64，128，strides=1，padding=1	56x56x128	73.856K	0.4624B
Conv4	1x1x128，64，strides=1，padding=0	56x56x64	8.256K	0.0514B
Conv5	3x3x64，128，strides=1，padding=1	56x56x128	73.856K	0.4624B
Max pool3	2x2，strides=2	28x28x128
Conv6	3x3x128，256，strides=1，padding=1	28x28x256	295.168K	0.4624B
Conv7	1x1x256，128，strides=1，padding=0	28x28x128	32.896K	0.0514B
Conv8	3x3x128，256，strides=1，padding=1	28x28x256	295.168K	0.4624B
Max pool4	2x2，strides=2	14x14x256
Conv9	3x3x256，512，strides=1，padding=1	14x14x512	1180.160K	0.4624B
Conv10	1x1x512，256，strides=1，padding=0	14x14x256	131.328K	0.0514B
Conv11	3x3x256，512，strides=1，padding=1	14x14x512	1180.160K	0.4624B
Conv12	1x1x512，256，strides=1，padding=0	14x14x256	131.328K	0.0514B
Conv13	3x3x256，512，strides=1，padding=1	14x14x512	1180.160K	0.4624B
Max pool5	2x2，strides=2	7x7x512
Conv14	3x3x512，1024，strides=1，padding=1	7x7x1024	4719.616K	0.4624B
Conv15	1x1x1024，512，strides=1，padding=0	7x7x512	524.800K	0.0514B
Conv16	3x3x512，1024，strides=1，padding=1	7x7x1024	4719.616K	0.4624B
Conv17	1x1x1024，512，strides=1，padding=0	7x7x512	524.800K	0.1028B
Conv18	3x3x512，1024，strides=1，padding=1	7x7x1024	4719.616K	0.4624B
Conv19	1x1x1024，1000，strides=1，padding=1	7x7x1000	1025.000M	0.1003B
Total			20.827M	5.582B