Faster-RCNN+VGG用自己的数据集训练模型
2016-09-06 21:46
260 查看
和ZF差不多，基本一样。不同的地方主要是网络模型的修改和训练结束后的修改。
1-6参考Faster-RCNN+ZF用自己的数据集训练模型
[plain] view
plain copy
input: "bbox_targets"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4*(类别数+1) (=21)
input_dim: 1
input_dim: 1
[plain] view
plain copy
input: "bbox_loss_weights"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4*(类别数+1)
input_dim: 1
input_dim: 1
[plain] view
plain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output: 5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
plain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output: 5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
plain copy
input: "bbox_targets"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4*(类别数+1)
input_dim: 1
input_dim: 1
[plain] view
plain copy
input: "bbox_loss_weights"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4* (类别数+1)
input_dim: 1
input_dim: 1
[plain] view
plain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output: 5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
plain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output:5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
！！！为防止与之前的模型搞混,训练前把output文件夹删除（或改个其他名），还要把imdb\cache中的文件删除（如果有的话）
[plain] view
plain copy
experiments\script_faster_rcnn_VOC2007_VGG16.m
[plain] view
plain copy
input: "data"
input_dim: 1
input_dim: 512
input_dim: 50
input_dim: 50
[plain] view
plain copy
layer {
bottom: "data"
bottom: "rois"
top: "pool5"
name: "roi_pool5"
type: "ROIPooling"
roi_pooling_param {
pooled_w: 7
pooled_h: 7
spatial_scale: 0.0625 # (1/16)
}
}
1-6参考Faster-RCNN+ZF用自己的数据集训练模型
7.网络的修改
（1）models\fast_rcnn_prototxts\vgg_16layers_fc6\train_val.prototxt
[plain] viewplain copy
input: "bbox_targets"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4*(类别数+1) (=21)
input_dim: 1
input_dim: 1
[plain] view
plain copy
input: "bbox_loss_weights"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4*(类别数+1)
input_dim: 1
input_dim: 1
[plain] view
plain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output: 5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
（2）models\fast_rcnn_prototxts\vgg_16layers_fc6\test.prototxt
[plain] viewplain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output: 5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
（3）models\fast_rcnn_prototxts\vgg_16layers_conv3_1\train_val.prototxt
[plain] viewplain copy
input: "bbox_targets"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4*(类别数+1)
input_dim: 1
input_dim: 1
[plain] view
plain copy
input: "bbox_loss_weights"
input_dim: 1 # to be changed on-the-fly to match num ROIs
input_dim: 20 # 4* (类别数+1)
input_dim: 1
input_dim: 1
[plain] view
plain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output: 5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
（4）models\fast_rcnn_prototxts\vgg_16layers_conv3_1\test.prototxt
[plain] viewplain copy
layer {
bottom: "fc7"
top: "cls_score"
name: "cls_score"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
type: "InnerProduct"
inner_product_param {
num_output:5 #类别数+1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
[plain] view
plain copy
layer {
bottom: "fc7"
top: "bbox_pred"
name: "bbox_pred"
type: "InnerProduct"
param {
lr_mult: 1.0
}
param {
lr_mult: 2.0
}
inner_product_param {
num_output: 20 #4*(类别数+1)
weight_filler {
type: "gaussian"
std: 0.001
}
bias_filler {
type: "constant"
value: 0
}
}
}
！！！为防止与之前的模型搞混,训练前把output文件夹删除（或改个其他名），还要把imdb\cache中的文件删除（如果有的话）
8.开始训练
[plain] viewplain copy
experiments\script_faster_rcnn_VOC2007_VGG16.m
9.训练完后
将relu5（包括relu5）前的层删除，并将roi_pool5的bottom改为data和rois。并且前面的input_dim:分别改为1,512,50,50，具体如下：[plain] view
plain copy
input: "data"
input_dim: 1
input_dim: 512
input_dim: 50
input_dim: 50
[plain] view
plain copy
layer {
bottom: "data"
bottom: "rois"
top: "pool5"
name: "roi_pool5"
type: "ROIPooling"
roi_pooling_param {
pooled_w: 7
pooled_h: 7
spatial_scale: 0.0625 # (1/16)
}
}
内容来自用户分享和网络整理，不保证内容的准确性，如有侵权内容，可联系管理员处理 
相关文章推荐
- Faster-RCNN+VGG用自己的数据集训练模型
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本and MATLAB版本）
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Matlab版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- [置顶] Faster-RCNN+ZF用自己的数据集训练模型(Matlab版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Matlab版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Python版本)
- Faster-RCNN+ZF用自己的数据集训练模型(Matlab版本)