| import tensorflow as tf |
| from ..ops import downsample as downsample_ops |
|
|
|
|
| def summarized_placeholder(name, prefix=None, key=tf.GraphKeys.SUMMARIES): |
| prefix = '' if not prefix else prefix + '/' |
| p = tf.placeholder(tf.float32, name=name) |
| tf.summary.scalar(prefix + name, p, collections=[key]) |
| return p |
|
|
|
|
| def resize_area(tensor, like): |
| _, h, w, _ = tf.unstack(tf.shape(like)) |
| return tf.stop_gradient(tf.image.resize_area(tensor, [h, w])) |
|
|
|
|
| def resize_bilinear(tensor, like): |
| _, h, w, _ = tf.unstack(tf.shape(like)) |
| return tf.stop_gradient(tf.image.resize_bilinear(tensor, [h, w])) |
|
|
| def downsample(tensor, num): |
| _,height, width,_ = tensor.shape.as_list() |
| if height%2==0 and width%2==0: |
| return downsample_ops(tensor, num) |
| else: |
| return tf.image.resize_area(tensor,tf.constant([int(height/num),int(width/num)])) |
| |
|
|
