From e372fe33ba47f81972b88562719296fe6a00a6c8 Mon Sep 17 00:00:00 2001
From: Piv <18462828+Piv200@users.noreply.github.com>
Date: Tue, 13 Jul 2021 20:32:45 +0930
Subject: [PATCH] Add per-pixel loss functions

---
 unsupervised/loss.py | 53 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100644 unsupervised/loss.py

diff --git a/unsupervised/loss.py b/unsupervised/loss.py
new file mode 100644
index 0000000..d62e5b1
--- /dev/null
+++ b/unsupervised/loss.py
@@ -0,0 +1,53 @@
+import tensorflow as tf
+
+
+def l1_loss(target_img, reprojected_img):
+    """
+    Calculates the l1 norm between the target and reprojected image
+
+    :param target_img: Tensor (batch, height, width, 3)
+    :param reprojected_img: Tensor, same shape as target_img
+    :return: The per-pixel l1 norm -> Tensor (batch, height, width, 1)
+    """
+    return tf.reduce_mean(tf.abs(target_img - reprojected_img), axis=3)
+
+
+def l2_loss(target_img, reprojected_img):
+    """
+    Calculates the l2 norm between the target and reprojected image
+
+    :param target_img: Tensor (batch, height, width, 3)
+    :param reprojected_img: Tensor, same shape as target_img
+    :return: The per-pixel l2 norm -> Tensor (batch, height, width, 1)
+    """
+    return tf.reduce_mean((target_img - reprojected_img) ** 2 ** (1 / 2), axis=3)
+
+
+def make_combined_ssim_l1_loss(ssim_weight: int = 0.85, other_loss_fn=l1_loss):
+    """
+    Create a loss function that will calculate ssim for the two images, and use the other_loss_fn to calculate the
+    per pixel loss
+    :param ssim_weight: Weighting that should be applied to SSIM weight vs l1 difference between target and
+        reprojected image
+    :param other_loss_fn: Function to combine with the ssim
+    :return: Function to calculate the per-pixel combined ssim with other loss function
+    """
+
+    def combined_ssim_loss(target_img, reprojected_img):
+        """
+        Calculates the per-pixel photometric reconstruction loss for each source image,
+        combined this with the SSIM between the reconstructed image and the actual image.
+
+        Calculates the following:
+        ssim_weight * SSIM(target_img, reprojected_img) + (1 - ssim_weight) * other_loss_fn(target_img - reprojected_img)
+
+        :param target_img: Tensor with shape (batch, height, width, 3) - current image we're training on
+        :param reprojected_img: Tensor with same shape as target_img, Reprojected from some source image that
+            should be as close as possible to the target image
+        :return: Per-pixel loss -> Tensor with shape (batch, height, width, 1), where height and width match target_img
+            height and width
+        """
+        ssim = tf.image.ssim(target_img, reprojected_img, axis=3, keepdim=True)
+        return ssim_weight * ssim + (1 - ssim_weight) * other_loss_fn(target_img, reprojected_img)
+
+    return combined_ssim_loss