版权声明:本文为博主原创文章,未经博主允许不得转载。 https://blog.csdn.net/laoxuan2011/article/details/52561707
caffe python wrapper
Load some useful libraries:
import os
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
List the system path
import sys
sys.path
Import caffe package
import caffe
Then, set the computation mode CPU or GPU
caffe.set_mode_cpu()
#or
#caffe.set_device(0)
#caffe.set_mode_gpu()
Create a network and load weights
net_model_file='./test.prototxt'
net_weights='./snapshot/train_iter_106000.caffemodel'
#load the model
net = caffe.Net(net_model_file,net_weights,caffe.TEST)
The net.params
are a list of [weight biases]
#weight in the conv1 layer
net.params['conv1'][0].data
#biases in the conv1 layer
net.params['conv1'][1].data