Darius Silkaitis

I got this running nnpcr.py This is an error message: `pickle.dump(obj, open(fileName + '.tmp', 'wb'), -1) SystemError: error return without exception set` I had 10k positive / 10k negative images....

Didn't get over 80% yet. Looking for better dataset. Basically I aim to train to recognize photos which are not suitable for advertisement networks. So even minor nudity is not...

Would it be smart to double image size? Would it be enough to change this constant? `IMG_SIZE = 128` to `IMG_SIZE = 256` Must positive and negative data samples be...

`def train(self, numIterations=1500):` Is this enough for batch of 60K of training data?

It really takes lot of ram. I could only run 40k sample on 32gb ram. Will try 80k on 64gb tomorrow. My model size always end up being 13mb, is...

By the way in new TensorFlow libarry you have to call `init_ops.zeros_initializer` like this, otherwise you will get error `init_ops.zeros_initializer()` I'll test with all those different parameters. Also I get...

Which line is for pre-outer layer?

``` x_image = tf.reshape(x, [-1, IMG_SIZE, IMG_SIZE, 3]) # 128 W_conv1 = tf.get_variable("W_conv1", shape=[3, 3, 3, 64], initializer=xavier()) b_conv1 = tf.get_variable('b_conv1', [1, 1, 1, 64]) h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) +...

No the problem I think is with what confident results softmax returns me. My guess is results are too dense. But I'm new to this field as you can see....

I'm glad to hear you like it. As I mentioned in package description it is Phapper package wrapper. Originally with captcha it tries to get your direct input directly from...