suyogduttjain / pixelobjectness Goto Github PK

Hi, I use this codes and run demo.py, there are 4-dimensional .mat files generated. But show_result.m　just process 3-dimensional .mat files, what is wrong with me??

thanks!!

When I run show_result.m file, there is a error:

Error using permute
ORDER must have at least N elements for an N-D array.

Error in show_results (line 23)
raw_result = permute(raw_result, [2 1 3]);

I tried to run pixelobjectness network in python.
following code is the code.

The output of following code is not the expected result.

I want to know what is the wrong code.

`
import numpy as np
from PIL import Image as PILImage
import caffe
_FOLDER = '/opt/pixelobjectness/'
MODEL_FILE = _FOLDER+'test.prototxt'
PRETRAINED = _FOLDER+'pixel_objectness.caffemodel'
net = caffe.Net(MODEL_FILE, PRETRAINED)

net.set_mode_gpu()
net.set_device(0)
net.set_phase_test()

mean_vec = np.array([104.008, 116.669, 122.675], dtype=np.float32)
reshaped_mean_vec = mean_vec.reshape(1, 1, 3);

net.set_raw_scale('data',255)
net.set_mean('data', reshaped_mean_vec)
net.set_channel_swap('data', (2, 1, 0))

IMAGE_FILE = '/opt/pixelobjectness/images/n01736375_3996.JPEG'
im = caffe.io.load_image(IMAGE_FILE)
im2 = net.preprocess('data',im)

net.blobs['data'].data[...] = im2

output = net.forward()
d = output['fc8_interp'].reshape(513,513,2,1)

import cPickle as pickle
with file("/root/workspace/images/a2.mat","w") as fout:
pickle.dump(output['fc8_interp'], fout)

`

Your demo is down

Hi,
I was trying to replicate the results presented in your paper and ran into some problems.
As far as I understand, you more or less follow the training procedure for DeepLab_LargeFOV network outlined here but with 2 classes instead of the original 21 and without the final CRF refinement layer (correct me if this is wrong).
I am training on the 10,582 images from the augmented PASCAL dataset, initialize the weights with VGG trained on ImageNet and use learning parameters as specified in the paper, however the results are not nearly as good. I haven't run the full set of tests yet, but on the example images results produced by my network are far worse than the published pixelobjectness model.
Also suspiciously, the loss fluctuates significantly and training for 10000 iterations takes about 3 hours on NVIDIA Titan X instead of 8 hours stated in the paper.
Below is the solver I've been using, could you please let me know if I'm missing something?

`lr_policy: "step"
gamma: 0.1
stepsize: 2000
base_lr: 0.001

display: 10
max_iter: 10000
momentum: 0.9
weight_decay: 0.0005`