search

WongKinYiu / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
GNU General Public License v3.0
13.37k stars 4.22k forks source link

question about E-ELAN in yolov7 #1624

Open momobiubiubiu opened 1 year ago

momobiubiubiu commented 1 year ago

in yolov7.yaml, the code about E-ELAN is as follows. I have a question about E-ELAN, why are their structures different? Both are E-ELAN?

[-1, 1, Conv, [256, 1, 1]], [-2, 1, Conv, [256, 1, 1]], [-1, 1, Conv, [256, 3, 1]], [-1, 1, Conv, [256, 3, 1]], [-1, 1, Conv, [256, 3, 1]], [-1, 1, Conv, [256, 3, 1]], [[-1, -3, -5, -6], 1, Concat, [1]], [-1, 1, Conv, [1024, 1, 1]]

2

[-1, 1, Conv, [256, 1, 1]], [-2, 1, Conv, [256, 1, 1]], [-1, 1, Conv, [128, 3, 1]], [-1, 1, Conv, [128, 3, 1]], [-1, 1, Conv, [128, 3, 1]], [-1, 1, Conv, [128, 3, 1]], [[-1, -2, -3, -4, -5, -6], 1, Concat, [1]], [-1, 1, Conv, [256, 1, 1]]

3

E-ELAN in paper as follows, why is it not the same as the structure in the code?

1
TusharChauhaan commented 1 year ago

[-1, 1, Conv, [256, 1, 1]

Could you please explain every term in the array.

nahidalam commented 1 year ago

@TusharChauhaan [256, 1, 1] part tells you the parameters of Convolution operations - channel out, kernel, stride respectively

changsubi commented 1 year ago

I want to know why it is different from proposed paper.

changsubi commented 1 year ago

sorry i found it yolov7-e6e.yaml