# Output Map
weight = weight_variable([1, 1, features, 1], stddev)
bias = bias_variable([1], -0.5)
conv = conv2d(in_node, weight, tf.constant(1.0))
pred_mask = tf.nn.sigmoid(conv + bias)
#pred_mask = tf.nn.relu(conv + bias)
# loss
#loss = tf.losses.mean_squared_error(pred_mask, crop(s_img,pred_mask))
#loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=s_, labels=s))
#loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=s_, labels=s))
loss = tf.reduce_mean(
tf.nn.sigmoid_cross_entropy_with_logits(
logits=conv + bias, labels=crop(s_img, pred_mask))) #sigmoid
#loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=pred_mask, labels=crop(s_img,pred_mask))) #relu
loss_sum = tf.summary.scalar("cross_entropy_early", loss)
loss_batch = tf.placeholder(tf.float32)
loss_batch_summary = tf.summary.scalar("cross_entropy_early", loss_batch)
#accuracy
#correct_prediction = tf.equal(tf.round(tf.clip_by_value(pred_mask, 0., 1.),),crop(s_img,pred_mask)) # relu
correct_prediction = tf.equal(tf.round(pred_mask, ),
crop(s_img, pred_mask)) # sigmoid
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float64))
accuracy_sum = tf.summary.scalar("accuracy_early", accuracy)
accuracy_batch = tf.placeholder(tf.float32)
convs.append((conv1, conv2))
# Output Map
weight = weight_variable([1, 1, features_root, 1], stddev)
bias = bias_variable([1], -0.5)
conv = conv2d(in_node, weight, tf.constant(1.0))
pred_mask = tf.nn.sigmoid(conv + bias)
# loss
#loss = tf.losses.mean_squared_error(pred_mask, crop(s_img,pred_mask))
#loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=s_, labels=s))
#loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=s_, labels=s))
loss = tf.reduce_mean(
tf.nn.sigmoid_cross_entropy_with_logits(
logits=conv + bias, labels=crop(s_img, pred_mask))) #sigmoid
#loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=pred_mask, labels=crop(s_img,pred_mask))) #relu
loss_sum = tf.summary.scalar("cross_entropy_unet", loss)
loss_batch = tf.placeholder(tf.float32)
loss_batch_summary = tf.summary.scalar("cross_entropy_unet", loss_batch)
#accuracy
#correct_prediction = tf.equal(tf.round(tf.clip_by_value(pred_mask, 0., 1.),),crop(s_img,pred_mask)) # relu
correct_prediction = tf.equal(tf.round(pred_mask), crop(s_img, pred_mask))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float64))
accuracy_sum = tf.summary.scalar("accuracy_unet", accuracy)
accuracy_batch = tf.placeholder(tf.float32)
accuracy_batch_summary = tf.summary.scalar("accuracy_unet", accuracy_batch)