caffe:Check failed: target_blobs.size() == source_layer.blobs_size() (2 vs. 1) Incompatible number of blobs for layer conv1

Question

I modify the FCN net and design a new net,in which I use two ImageData Layer as input param and hope the net produces a picture as output. here is the train_val.prototxt and the deploy.prototxt

the original picture and the label are both gray scale pics and sizes are 224*224. I've trained a caffemodel and use infer.py to use the caffemodel to do a segmentation,but meet the error:

 F0505 06:15:08.072602 30713 net.cpp:767] Check failed: target_blobs.size() == source_layer.blobs_size() (2 vs. 1) Incompatible number of blobs for layer conv1

here is the infer.py file:

import numpy as np
from PIL import Image
caffe_root = '/home/zhaimo/' 
import sys
sys.path.insert(0, caffe_root + 'caffe-master/python')

import caffe
im = Image.open('/home/zhaimo/fcn-master/data/vessel/test/13.png')
in_ = np.array(im, dtype=np.float32)
#in_ = in_[:,:,::-1]
#in_ -= np.array((104.00698793,116.66876762,122.67891434))
#in_ = in_.transpose((2,0,1))


net = caffe.Net('/home/zhaimo/fcn-master/mo/deploy.prototxt', '/home/zhaimo/fcn-master/mo/snapshot/train/_iter_200000.caffemodel', caffe.TEST)
net.blobs['data'].reshape(1, *in_.shape)
net.blobs['data'].data[...] = in_
net.forward()
out = net.blobs['score'].data[0].argmax(axis=0)

plt.axis('off')
plt.savefig('/home/zhaimo/fcn-master/mo/result/13.png')

how to solve this problem?

nomem · Accepted Answer · 2017-05-06 04:18:47Z

2

The problem is with your bias term in conv1. In your train.prototxt it is set to false. But in your deploy.prototxt it is not and by default that is true. That is why weight loader is looking for two blobs.

answered May 6, 2017 at 4:18

nomem

1,5884 gold badges17 silver badges35 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

StalkerMuse Over a year ago

here is another error:

F0508 03:20:57.777287 22515 base_conv_layer.cpp:189] Check failed: bottom[0]->num_axes() == first_spatial_axis + num_spatial_axes_ (3 vs. 4) bottom num_axes may not change.

I guess it is because the dim in input_param should be 1,1,224,224,so I change it but meet another error:

Cannot copy param 0 weights from layer 'conv1'; shape mismatch.  Source param shape is 64 3 7 7 (9408); target param shape is 64 1 7 7 (3136). To learn this layer's parameters from scratch rather than copying from a saved net, rename the layer

how to solve this problem,please?

StalkerMuse Over a year ago

btw I'm sure my train and test data are both 8bit gray scale pics.

nomem Over a year ago

This 2nd issue is with the number of channels (rgb vs gray). Your train and test both should have same number of channels. So do appropriate conversion based on your need.

StalkerMuse Over a year ago

thanks a lot.I changed the deploy.prototxt and trainval.prototxt,but still can not solve the first error.I still can not figure out the problem because all pics' channels are the same.

Collectives™ on Stack Overflow

caffe:Check failed: target_blobs.size() == source_layer.blobs_size() (2 vs. 1) Incompatible number of blobs for layer conv1

1 Answer 1

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related