I have changed the backbone of scaled-yolov4 p5,add the Switchable Atrous Convolution ,which come from the DetectoRs,in the backbone,but it doesn't work well, the mAP in my dataset from 43.6 down to 17.8...I had expected it to improve the value of mAP.
This is my config:
[from, number, module, args]
[[-1, 1, Conv, [32, 3, 1]], # 0
[-1, 1, Conv, [64, 3, 2]], # 1-P1/2
[-1, 1, BottleneckCSP, [64]],
[-1, 1, Conv, [128, 3, 2]], # 3-P2/4
[-1, 3, BottleneckCSP, [128]],
[-1, 1, Conv, [256, 3, 2]], # 5-P3/8
[-1, 15, BottleneckCSP, [256]],
[-1, 1, Conv, [512, 3, 2]], # 7-P4/16
[-1, 15, BottleneckCSP, [512]],
[-1, 1, SACConv, [1024, 3, 2]], # 9-P5/32
[-1, 7, BottleneckCSP, [1024]], # 10
]
the SACConv use SAConv2d to replace nn.Conv2d in the class Conv .
Do you know why? please tell me .thanks very much.
I have changed the backbone of scaled-yolov4 p5,add the Switchable Atrous Convolution ,which come from the DetectoRs,in the backbone,but it doesn't work well, the mAP in my dataset from 43.6 down to 17.8...I had expected it to improve the value of mAP. This is my config:
[from, number, module, args]
[[-1, 1, Conv, [32, 3, 1]], # 0 [-1, 1, Conv, [64, 3, 2]], # 1-P1/2 [-1, 1, BottleneckCSP, [64]], [-1, 1, Conv, [128, 3, 2]], # 3-P2/4 [-1, 3, BottleneckCSP, [128]], [-1, 1, Conv, [256, 3, 2]], # 5-P3/8 [-1, 15, BottleneckCSP, [256]], [-1, 1, Conv, [512, 3, 2]], # 7-P4/16 [-1, 15, BottleneckCSP, [512]], [-1, 1, SACConv, [1024, 3, 2]], # 9-P5/32 [-1, 7, BottleneckCSP, [1024]], # 10 ] the SACConv use SAConv2d to replace nn.Conv2d in the class Conv . Do you know why? please tell me .thanks very much.