Re-export all nuScenes tutorials (nutonomy#478)

* Outputs for map_expansion_tutorial.ipynb

* Outputs for nuimages_tutorial.ipynb

* Outputs for nuscenes_lidarseg_tutorial.ipynb

* Outputs for nuscenes_tutorial.ipynb

* Outputs for prediction_tutorial.ipynb

* Fix typo in LidarSegEval

* Compare with gt if available when validating no. of points in preds

* Allow print_all_message_stats in can bus to print partial only

* Outputs for can_bus_tutorial.ipynb

* Clear output from tutorials

python-sdk/nuscenes/can_bus/can_bus_api.py

@@ -62,16 +62,23 @@ def __init__(self,
         self.all_messages = self.can_messages + self.derived_messages
 
     def print_all_message_stats(self,
-                                scene_name: str) -> None:
+                                scene_name: str,
+                                print_full: bool = False) -> None:
         """
         Prints the meta stats for each CAN message type of a particular scene.
         :param scene_name: The name of the scene, e.g. scene-0001.
+        :param print_full: Whether to show all stats for all message types in the scene.
         """
         all_messages = {}
         for message_name in self.can_messages:
             messages = self.get_messages(scene_name, 'meta')
             all_messages[message_name] = messages
-        print(json.dumps(all_messages, indent=2))
+
+        if print_full:
+            print(json.dumps(all_messages, indent=2))
+        else:
+            partial_messages = {message: list(stats.keys()) for message, stats in all_messages.items()}
+            print(json.dumps(partial_messages, indent=2))
 
     def print_message_stats(self,
                             scene_name: str,

python-sdk/nuscenes/eval/lidarseg/evaluate.py

@@ -59,7 +59,7 @@ def __init__(self,
         self.eval_set = eval_set
         self.verbose = verbose
 
-        self.mapper = LidarsegClassMapper(nusc_)
+        self.mapper = LidarsegClassMapper(self.nusc)
         self.ignore_idx = self.mapper.ignore_class['index']
         self.id2name = {idx: name for name, idx in self.mapper.coarse_name_2_coarse_idx_mapping.items()}
         self.num_classes = len(self.mapper.coarse_name_2_coarse_idx_mapping)

python-sdk/nuscenes/eval/lidarseg/validate_submission.py

@@ -82,16 +82,24 @@ def validate_submission(nusc: NuScenes, results_folder: str, eval_set: str, verb
             'Error: The prediction .bin file {} does not exist.'.format(lidarseg_pred_filename)
         lidarseg_pred = np.fromfile(lidarseg_pred_filename, dtype=np.uint8)
 
-        # Check number of points in the point cloud.
-        pointsensor = nusc.get('sample_data', sd_token)
-        pcl_path = os.path.join(nusc.dataroot, pointsensor['filename'])
-        pc = LidarPointCloud.from_file(pcl_path)
-        points = pc.points
-
-        assert points.shape[1] == len(lidarseg_pred), \
+        # Check number of predictions for the point cloud.
+        if len(nusc.lidarseg) > 0:  # If ground truth exists, compare the no. of predictions with that of ground truth.
+            lidarseg_label_filename = os.path.join(results_bin_folder, sd_token + '_lidarseg.bin')
+            assert os.path.exists(lidarseg_pred_filename), \
+                'Error: The ground truth .bin file {} does not exist.'.format(lidarseg_label_filename)
+            lidarseg_label = np.fromfile(lidarseg_label_filename, dtype=np.uint8)
+            num_points = len(lidarseg_label)
+        else:  # If no ground truth is available, compare the no. of predictions with that of points in a point cloud.
+            pointsensor = nusc.get('sample_data', sd_token)
+            pcl_path = os.path.join(nusc.dataroot, pointsensor['filename'])
+            pc = LidarPointCloud.from_file(pcl_path)
+            points = pc.points
+            num_points = points.shape[1]
+
+        assert num_points == len(lidarseg_pred), \
             'Error: There are {} predictions for lidar sample data token {} ' \
             'but there are only {} points in the point cloud.'\
-            .format(sd_token, len(lidarseg_pred), points.shape[1])
+            .format(sd_token, len(lidarseg_pred), num_points)
 
         assert all((lidarseg_pred > 0) & (lidarseg_pred < num_classes)), \
             "Error: Array for predictions in {} must be between 1 and {} (inclusive)."\

python-sdk/tutorials/can_bus_tutorial.ipynb

@@ -208,7 +208,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.6"
+   "version": "3.7.7"
   },
   "pycharm": {
    "stem_cell": {

python-sdk/tutorials/prediction_tutorial.ipynb

@@ -5,7 +5,7 @@
    "metadata": {},
    "source": [
     "# nuScenes prediction tutorial\n",
-    "<img src=\"trajectory.gif\" width=\"300\" align=\"left\">"
+    "<img src=\"https://www.nuscenes.org/public/tutorials/trajectory.gif\" width=\"300\" align=\"left\">"
    ]
   },
   {