checkpoint conclusions

01_CARE/care_solution.ipynb

@@ -514,9 +514,26 @@
    "metadata": {},
    "source": [
     "<div class=\"alert alert-block alert-success\"><h1>Checkpoint 1: Data</h1>\n",
+    "\n",
+    "In this section, we prepared paired training data. \n",
+    "The steps were:\n",
+    "1) Loading the images.\n",
+    "2) Cropping them into patches.\n",
+    "3) Checking the patches visually.\n",
+    "4) Creating an instance of a pytorch dataset and dataloader.\n",
+    "\n",
+    "You'll see a similar preparation procedure followed for most deep learning vision tasks.\n",
+    "\n",
+    "Next, we'll use this data to train a denoising model.\n",
     "</div>\n",
     "\n",
-    "<hr style=\"height:2px;\">\n",
+    "<hr style=\"height:2px;\">\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
     "\n",
     "## Part 2: Training the model\n",
     "\n",
@@ -773,9 +790,25 @@
    "metadata": {},
    "source": [
     "<div class=\"alert alert-block alert-success\"><h1>Checkpoint 2: Training</h1>\n",
+    "\n",
+    "In this section, we created and trained a UNet for denoising.\n",
+    "We:\n",
+    "1) Instantiated the model with random weights.\n",
+    "2) Chose a loss function to compare the output image to the ground truth clean image.\n",
+    "3) Chose an optimizer to minimize that loss function.\n",
+    "4) Trained the model with this optimizer.\n",
+    "5) Examined the training and validation loss curves to see how well our model trained.\n",
+    "\n",
+    "Next, we'll load a test set of noisy images and see how well our model denoises them.\n",
     "</div>\n",
     "\n",
-    "<hr style=\"height:2px;\">\n",
+    "<hr style=\"height:2px;\">\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
     "\n",
     "## Part 3: Predicting on the test dataset\n"
    ]
@@ -923,8 +956,21 @@
    "metadata": {},
    "source": [
     "<div class=\"alert alert-block alert-success\"><h1>Checkpoint 3: Predicting</h1>\n",
+    "\n",
+    "In this section, we evaluated the performance of our denoiser.\n",
+    "We:\n",
+    "1) Created a CAREDataset and Dataloader for a prediction loop.\n",
+    "2) Ran a prediction loop on the test data.\n",
+    "3) Examined the outputs.\n",
+    "\n",
+    "This notebook has shown how matched pairs of noisy and clean images can train a UNet to denoise, but what if we don't have any clean images? In the next notebook, we'll try Noise2Void, a method for training a UNet to denoise with only noisy images.\n",
     "</div>"
    ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": []
   }
  ],
  "metadata": {