[fix]: fix demo results bug caused by batch inference support (#116)

mmdet3d/apis/inference.py

@@ -99,7 +99,7 @@ def show_result_meshlab(data, result, out_dir):
 
     assert out_dir is not None, 'Expect out_dir, got none.'
 
-    pred_bboxes = result['boxes_3d'].tensor.numpy()
+    pred_bboxes = result[0]['boxes_3d'].tensor.numpy()
     # for now we convert points into depth mode
     if data['img_metas'][0][0]['box_mode_3d'] != Box3DMode.DEPTH:
         points = points[..., [1, 0, 2]]