yolov3转换 Upsample Layer出错

[zgq91]

hi:
感谢开源 我在转成darknet官方的yolov3.cfg yolov3.weights时出错:Check failed: ExactNumBottomBlobs() == bottom.size() (2 vs. 1) Upsample Layer takes 2 bottom blob(s) as input.环境windows10 python3.6

[ChenYingpeng]

Please check your upsample layer which botton blobs size must be 1. Such as
`

layer {
bottom: "layer85-conv"
top: "layer86-upsample"
name: "layer86-upsample"
type: "Upsample"
upsample_param {
scale: 2
}
}
`

[yulinhuyang]

layer {
bottom: "layer118-conv"
top: "layer119-upsample"
name: "layer119-upsample"
type: "Upsample"
upsample_param {
scale: 2
}
} 这是我的转换结果，在一些嵌入式AI环境仿真也报同样错误，没看明白为啥。

[yulinhuyang]

看到网上又用deconv替换upsample的。不知道能否解决

[ChenYingpeng]

Please post more details about your error log.