visualizer and notebook fixes

Signed-off-by: Igor Davidyuk <igor.davidyuk@intel.com>

geti_sdk/prediction_visualization/visualizer.py

@@ -86,7 +86,7 @@ def draw(
         if confidence_threshold is not None:
             annotation = annotation.filter_by_confidence(confidence_threshold)
         result = self.shape_drawer.draw(
-            image, annotation, labels=[], fill_shapes=fill_shapes
+            image.copy(), annotation, labels=[], fill_shapes=fill_shapes
         )
         return result
 

notebooks/003_upload_and_predict_image.ipynb

@@ -214,9 +214,10 @@
     "# To visualise the image, we have to retrieve the pixel data from the platform using the `image.get_data` method. The actual pixel data is\n",
     "# downloaded and cached only on the first call to this method\n",
     "image.get_data(geti.session)\n",
+    "numpy_image = image.numpy\n",
     "\n",
     "visualizer = Visualizer()\n",
-    "image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
+    "image_rgb = cv2.cvtColor(numpy_image, cv2.COLOR_BGR2RGB)\n",
     "result = visualizer.draw(image_rgb, prediction)\n",
     "visualizer.show_in_notebook(result)"
    ]

notebooks/008_deploy_project.ipynb

@@ -95,7 +95,9 @@
    "id": "ec4c279c-c6f1-4663-8c41-f96431212252",
    "metadata": {},
    "source": [
-    "Once we are sure that the project has trained models for each task, we can create the deployment in the cell below."
+    "Once we are sure that the project has trained models for each task, we can create the deployment in the cell below.\n",
+    "\n",
+    "Note the `enable_explainable_ai` argument. If set to `True`, the deployment will include the necessary artifacts to run the Explainable AI (XAI) service. This will allow us to generate explanations for the predictions made by the models in the deployment."
    ]
   },
   {
@@ -105,7 +107,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "deployment = geti.deploy_project(project_name=PROJECT_NAME)"
+    "deployment = geti.deploy_project(project_name=PROJECT_NAME, enable_explainable_ai=True)"
    ]
   },
   {
@@ -191,6 +193,27 @@
     "visualizer.show_in_notebook(result)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f954741c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "t_start = time.time()\n",
+    "prediction_with_saliency_map = deployment.explain(numpy_rgb)\n",
+    "t_elapsed = time.time() - t_start\n",
+    "\n",
+    "print(\n",
+    "    f\"Running local inference with XAI on image took {t_elapsed*1000:.2f} milliseconds\"\n",
+    ")\n",
+    "\n",
+    "result = visualizer.explain_label(\n",
+    "    numpy_rgb, prediction_with_saliency_map, label_name=\"dog\"\n",
+    ")\n",
+    "visualizer.show_in_notebook(result)"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,