learn caffe 1

最新推荐文章于 2022-11-26 15:05:51 发布

mathilde27

最新推荐文章于 2022-11-26 15:05:51 发布

阅读量335

点赞数

分类专栏：深度学习

本文链接：https://blog.youkuaiyun.com/Chunying27/article/details/54586562

版权

深度学习专栏收录该内容

27 篇文章

订阅专栏

本文详细介绍了如何使用Caffe框架加载预训练的深度学习模型，并展示了模型中各层的结构与参数形状，适用于理解模型结构及进行迁移学习。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

load Net

caffe.set_mode_cpu()

model_def = 'deploy.prototxt'
model_weights ='bvlc.caffemodel'
net = caffe.Net(model_def,      # defines the structure of the model
                model_weights,  # contains the trained weights
                caffe.TEST)     # use test mode

net.blobs

Each layer

for layer_name,blob in net.blobs.items():
    print(layer_name+"\t"+str(blob.data.shape))

data    (5, 3, 227, 227)
conv1   (5, 96, 55, 55)
pool1   (5, 96, 27, 27)
norm1   (5, 96, 27, 27)
conv2   (5, 256, 27, 27)
pool2   (5, 256, 13, 13)
norm2   (5, 256, 13, 13)
conv3   (5, 384, 13, 13)
conv4   (5, 384, 13, 13)
conv5   (5, 256, 13, 13)
pool5   (5, 256, 6, 6)
fc6 (5, 4096)
fc7 (5, 4096)
fc8 (5, 1000)
prob    (5, 1000)

net.params



for layer_name,param in net.params.items():
    print(layer_name,'\t',param[0].data.shape,param[1].data.shape)

conv1    (96, 3, 11, 11) (96,)
conv2    (256, 48, 5, 5) (256,)
conv3    (384, 256, 3, 3) (384,)
conv4    (384, 192, 3, 3) (384,)
conv5    (256, 192, 3, 3) (256,)
fc6      (4096, 9216) (4096,)
fc7      (4096, 4096) (4096,)
fc8      (1000, 4096) (1000,)