add contiguous() after transpose() at pointrcnn_head (open-mmlab#373)

* bugfixed: ignore empty boxes in visualization

* add contiguous() after transpose() at pointrcnn_head.py

pcdet/models/dense_heads/anchor_head_multi.py

@@ -150,7 +150,7 @@ def forward(self, spatial_features_2d):
 
 class AnchorHeadMulti(AnchorHeadTemplate):
     def __init__(self, model_cfg, input_channels, num_class, class_names, grid_size, point_cloud_range,
-                 predict_boxes_when_training=True):
+                 predict_boxes_when_training=True, **kwargs):
         super().__init__(
             model_cfg=model_cfg, num_class=num_class, class_names=class_names, grid_size=grid_size,
             point_cloud_range=point_cloud_range, predict_boxes_when_training=predict_boxes_when_training

pcdet/models/dense_heads/anchor_head_single.py

@@ -6,7 +6,7 @@
 
 class AnchorHeadSingle(AnchorHeadTemplate):
     def __init__(self, model_cfg, input_channels, num_class, class_names, grid_size, point_cloud_range,
-                 predict_boxes_when_training=True):
+                 predict_boxes_when_training=True, **kwargs):
         super().__init__(
             model_cfg=model_cfg, num_class=num_class, class_names=class_names, grid_size=grid_size, point_cloud_range=point_cloud_range,
             predict_boxes_when_training=predict_boxes_when_training

pcdet/models/roi_heads/pointrcnn_head.py

@@ -147,7 +147,7 @@ def forward(self, batch_dict):
 
         pooled_features = self.roipool3d_gpu(batch_dict)  # (total_rois, num_sampled_points, 3 + C)
 
-        xyz_input = pooled_features[..., 0:self.num_prefix_channels].transpose(1, 2).unsqueeze(dim=3)
+        xyz_input = pooled_features[..., 0:self.num_prefix_channels].transpose(1, 2).unsqueeze(dim=3).contiguous()
         xyz_features = self.xyz_up_layer(xyz_input)
         point_features = pooled_features[..., self.num_prefix_channels:].transpose(1, 2).unsqueeze(dim=3)
         merged_features = torch.cat((xyz_features, point_features), dim=1)