 Net Surgery (卷积，全连接的weights和biases的参数存储方式理解))
★ FEATURED ARTICLE importnumpyasnpimportmatplotlib.pyplotaspltimportsys%matplotlib inlineimportcaffe caffe_example_pathyour caffe example pathsys.path.append(caffe_example_path)#configure plottingplt.rcParams[figure.figsize](10,10)plt.rcParams[image.interpolation]nearestplt.rcParams[image.cmap]graycaffe.set_mode_cpu()netcaffe.Net(caffe_example_pathnet_surgery/conv.prototxt,caffe.TEST)print(blobs {}\n params {}.format(net.blobs.keys(),net.params.keys()))blobs odict_keys([data, conv]) params odict_keys([conv])imnp.array(caffe.io.load_image(caffe_example_pathimages/cat gray.jpg,colorFalse)).squeeze()plt.title(original image)plt.imshow(im)plt.axis(off)/home/felaim/anaconda3/envs/tensorflow-gpu/lib/python3.7/site-packages/skimage/io/_io.py:48: UserWarning: as_grey has been deprecated in favor of as_gray warn(as_grey has been deprecated in favor of as_gray) (-0.5, 479.5, 359.5, -0.5)print(im.shape: ,im.shape)im_inputim[np.newaxis,np.newaxis,:,:]print(im_input.shape: ,im_input.shape)net.blobs[data].reshape(*im_input.shape)print(reshape im_input.shape: ,im_input.shape)net.blobs[data].data[...]im_inputim.shape: (360, 480) im_input.shape: (1, 1, 360, 480) reshape im_input.shape: (1, 1, 360, 480)defshow_filters(net):net.forward()plt.figure()filt_min,filt_maxnet.blobs[conv].data.min(),net.blobs[conv].data.max()foriinrange(3):plt.subplot(1,4,i2)plt.title(filter #{} output.format(i))plt.imshow(net.blobs[conv].data[0,i],vminfilt_min,vmaxfilt_max)plt.tight_layout()plt.axis(off)# filter the image with initialshow_filters(net)# pick first filter outputconv0net.blobs[conv].data[0,0]print(pre-surgery output mean {:.2f}.format(conv0.mean()))pre-surgery output mean 0.00# set first filter bias to 1net.params[conv][1].data[0]1net.forward()print(post-surgery output mean {:.2f}.format(conv0.mean()))post-surgery output mean 1.00ksizenet.params[conv][0].data.shape[2:]print(ksize: ,ksize)# make Gaussian blursigma1y,xnp.mgrid[-ksize[0]//21:ksize[0]//21,-ksize[1]//21:ksize[1]//21]print(y: ,y, \n x: ,x)ksize: (5, 5) y: [[-2 -2 -2 -2 -2] [-1 -1 -1 -1 -1] [ 0 0 0 0 0] [ 1 1 1 1 1] [ 2 2 2 2 2]] x: [[-2 -1 0 1 2] [-2 -1 0 1 2] [-2 -1 0 1 2] [-2 -1 0 1 2] [-2 -1 0 1 2]]gnp.exp(-((x**2y**2)/(2.0*sigma**2)))print(g:,g)g: [[0.01831564 0.082085 0.13533528 0.082085 0.01831564] [0.082085 0.36787944 0.60653066 0.36787944 0.082085 ] [0.13533528 0.60653066 1. 0.60653066 0.13533528] [0.082085 0.36787944 0.60653066 0.36787944 0.082085 ] [0.01831564 0.082085 0.13533528 0.082085 0.01831564]]gaussian(g/g.sum()).astype(np.float32)net.params[conv][0].data[0]gaussian# make sobel operator for edge detectionnet.params[conv][0].data[1:]0sobelnp.array((-1,-2,-1,0,0,0,1,2,1),dtypenp.float32).reshape((3,3))net.params[conv][0].data[1,0,1:-1,1:-1]sobel# horizontalnet.params[conv][0].data[2,0,1:-1,1:-1]sobel.T# verticalshow_filters(net)!diff/home/felaim/Documents/software/caffe-master/examples/net_surgery/bvlc_caffenet_full_conv.prototxt/home/felaim/Documents/software/caffe-master/models/bvlc_reference_caffenet/deploy.prototxt1,2c1 # Fully convolutional network version of CaffeNet. name: CaffeNetConv --- name: CaffeNet 7,11c6 input_param { # initial shape for a fully convolutional network: # the shape can be set for each input by reshape. shape: { dim: 1 dim: 3 dim: 451 dim: 451 } } --- input_param { shape: { dim: 10 dim: 3 dim: 227 dim: 227 } } 157,158c152,153 name: fc6-conv type: Convolution --- name: fc6 type: InnerProduct 160,161c155,156 top: fc6-conv convolution_param { --- top: fc6 inner_product_param { 163d157 kernel_size: 6 169,170c163,164 bottom: fc6-conv top: fc6-conv --- bottom: fc6 top: fc6 175,176c169,170 bottom: fc6-conv top: fc6-conv --- bottom: fc6 top: fc6 182,186c176,180 name: fc7-conv type: Convolution bottom: fc6-conv top: fc7-conv convolution_param { --- name: fc7 type: InnerProduct bottom: fc6 top: fc7 inner_product_param { 188d181 kernel_size: 1 194,195c187,188 bottom: fc7-conv top: fc7-conv --- bottom: fc7 top: fc7 200,201c193,194 bottom: fc7-conv top: fc7-conv --- bottom: fc7 top: fc7 207,211c200,204 name: fc8-conv type: Convolution bottom: fc7-conv top: fc8-conv convolution_param { --- name: fc8 type: InnerProduct bottom: fc7 top: fc8 inner_product_param { 213d205 kernel_size: 1 219c211 bottom: fc8-conv --- bottom: fc8# Load the original network and extract the fully connected layers parameterscaffe_model_pathpath to caffe/modelsnetcaffe.Net(caffe_model_path/bvlc_reference_caffenet/deploy.prototxt,caffe_model_path/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel,caffe.TEST)params[fc6,fc7,fc8]fc_params{pr:(net.params[pr][0].data,net.params[pr][1].data)forprinparams}forfcinparams:print({} weights are {} dimensional and biases are {} dimensional.format(fc,fc_params[fc][0].shape,fc_params[fc][1].shape))fc6 weights are (4096, 9216) dimensional and biases are (4096,) dimensional fc7 weights are (4096, 4096) dimensional and biases are (4096,) dimensional fc8 weights are (1000, 4096) dimensional and biases are (1000,) dimensionalnet_full_convcaffe.Net(caffe_example_pathnet_surgery/bvlc_caffenet_full_conv.prototxt,caffe_model_path/bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel,caffe.TEST)params_full_conv[fc6-conv,fc7-conv,fc8-conv]# conv_params {name:(weights, biases)}conv_params{pr:(net_full_conv.params[pr][0].data,net_full_conv.params[pr][1].data)forprinparams_fully_conv}forconvinparams_full_conv:print({} weights are {} dimensional and biases are {} dimensional.format(conv,conv_params[conv][0].shape,conv_params[conv][1].shape))fc6-conv weights are (4096, 256, 6, 6) dimensional and biases are (4096,) dimensional fc7-conv weights are (4096, 4096, 1, 1) dimensional and biases are (4096,) dimensional fc8-conv weights are (1000, 4096, 1, 1) dimensional and biases are (1000,) dimensionalforpr,pr_convinzip(params,params_full_conv):conv_params[pr_conv][0].flatfc_params[pr][0].flat# flat unrolls the arraysconv_params[pr_conv][1][...]fc_params[pr][1]net_full_conv.save(caffe_example_pathnet_surgery/bvlc_caffenet_full_conv.caffemodel)importnumpyasnpimportmatplotlib.pyplotasplt%matplotlib inline#load input and configure prepeocessingimcaffe.io.load_image(caffe_example_pathimages/cat.jpg)transformercaffe.io.Transformer({data:net_full_conv.blobs[data].data.shape})transformer.set_mean(data,np.load(../python/caffe/imagenet/ilsvrc_2012_mean.npy).mean(1).mean(1))transformer.set_transpose(data,(2,0,1))transformer.set_channel_swap(data,(2,1,0))transformer.set_raw_scale(data,255.0)# make classification map by forward and print prediction indices at each locationoutnet_full_conv.forward_all(datanp.asarray([transformer.preprocess(data,im)]))print(out[prob][0].argmax(axis0))/home/felaim/anaconda3/envs/tensorflow-gpu/lib/python3.7/site-packages/skimage/io/_io.py:48: UserWarning: as_grey has been deprecated in favor of as_gray warn(as_grey has been deprecated in favor of as_gray) [[282 282 281 281 281 281 277 282] [281 283 283 281 281 281 281 282] [283 283 283 283 283 283 287 282] [283 283 283 281 283 283 283 259] [283 283 283 283 283 283 283 259] [283 283 283 283 283 283 259 259] [283 283 283 283 259 259 259 277] [335 335 283 259 263 263 263 277]]# show net input and confidence map (probability of the top prediction at each location)plt.subplot(1,2,1)plt.imshow(transformer.deprocess(data,net_full_conv.blobs[data].data[0]))plt.subplot(1,2,2)plt.imshow(out[prob][0,281])matplotlib.image.AxesImage at 0x7f40a17a8208理解部分全连接的参数的存储方式权重尺寸是先输出尺寸后输入尺寸而偏差尺寸是输出尺寸。权重存储位置 fc_params[fc][0] 偏差存储位置 fc_params[fc][1]卷积的参数存储方式卷积权重按输出×输入×高度×宽度尺寸排列。权重存储位置conv_params[conv][0] 偏差存储位置conv_params[conv][1]参考地址https://nbviewer.jupyter.org/github/BVLC/caffe/blob/master/examples/net_surgery.ipynb
阅读完成 · 觉得有帮助？
