Missing batch normalization

[ohuet74]

Hello
I don't understand your prototxt of yolo.
You have removed all the batch normalization layer and the scale after convolution
For me a convolution is
layer {
name: "conv2d_1"
type: "Convolution"
bottom: "data"
top: "conv2d_1"
convolution_param {
num_output: 32
kernel_size: 3
stride: 1
pad : 1
}
}
layer {
name: "batch_normalization_1"
type: "BatchNorm"
bottom: "conv2d_1"
top: "conv2d_1"
batch_norm_param {
eps : 0.000001
}
}
layer {
name: "scale_1"
type: "Scale"
bottom: "conv2d_1"
top: "conv2d_1"
scale_param {
bias_term : true
}
}
layer {
name: "relu_1"
type: "ReLU"
bottom: "conv2d_1"
top: "conv2d_1"
relu_param {
negative_slope: 0.1
}
}
the batch_normalization take training parameter if y remove it for me it's doesn't work.

[convolutional]
batch_normalize=1
filters=64
size=7
stride=2
pad=1
activation=leaky