caffe中ImageData layer的圖像增強操作
mirror
mirror:ture
代表隨機的左右翻轉。It is random left-right flipping, a common operating when training models.
contrast_brightness_adjustment
開啓或者禁止對比度調節,默認禁止(false)
contrast_brightness_adjustment:true
min_side_min
min_side_min and min_side_max are added for random cropping while keeping the aspect ratio。 as mentioned in “Deep Residual Learning for Image Recognition”(http://arxiv.org/abs/1512.03385)
min_side_min:224
使用了min_side_min和min_side_max就不需要再在image_data_param中設置new_height和new_width兩個參數。在這裏圖片將被隨機resize到這個區間內
min_side_max
min_side_max:256
crop_size
crop_size:224
在caffe中,如果定義了crop_size,那麼在train時會對大於crop_size的圖片進行隨機裁剪,而在test時只是截取中間部分max_rotation_angle
圖片最大的旋轉角度,默認爲0
max_rotation_angle:15
min_contrast
最小的對比度乘子(min alpha),默認0.8
max_contrast
最大對比度乘子(max alpha),默認1.2
max_smooth
最大平滑乘子,默認6
進行高斯平滑apply_probability
每個操作被執行的概率,默認爲0.5
max_color_shift
在RGB軸上最大的色彩偏移
max_color_shift:20
mean_value:
BGR順序的均值
debug_params
使能或禁止打印操作參數,默認禁止
debug_params:false
min_side
resize & crop 保持縱橫比,默認0,disabled
max_brightness_shift
max brightness shift in positive and negative directions (beta), default 5;
smooth_filtering
enable/disable smooth filterion, default false;
例子
layer {
name: "in_shop"
type: "ImageData"
top: "data"
top: "label"
include{
phase: TRAIN
}
transform_param {
mirror: true
contrast_brightness_adjustment: true
min_side_min: 224
min_side_max: 256
crop_size: 224
max_rotation_angle: 15
min_contrast: 0.8
max_contrast: 1.2
max_smooth: 6
apply_probability: 0.5
max_color_shift: 20
mean_value: 104
mean_value: 117
mean_value: 123
debug_params: false
}
image_data_param {
source: "/export/home/dyh/workspace/circle_k/for_douyuhao/all-sample.txt"
batch_size: 128
new_height: 256
new_width: 256
shuffle: true
root_folder: "/export/home/dyh/workspace/circle_k/for_douyuhao/all-images/"
}
}
參考:https://github.com/twtygqyy/caffe-augmentation/blob/master/README.md