Join us at TensorFlow World, Oct 28-31. Use code TF20 for 20% off select passes. Register now


View source on GitHub

A reshaped summary image.


  • tf.contrib.gan.eval.eval_utils.image_reshaper

Returns an image that will contain all elements in the list and will be laid out in a nearly-square tiling pattern (e.g. 11 images will lead to a 3x4 tiled image).


  • images: Image data to summarize. Can be an RGB or grayscale image, a list of such images, or a set of RGB images concatenated along the depth dimension. The shape of each image is assumed to be [batch_size, height, width, depth].
  • num_cols: (Optional) If provided, this is the number of columns in the final output image grid. Otherwise, the number of columns is determined by the number of images.


A summary image matching the input with automatic tiling if needed. Output shape is [1, height, width, channels].