Fix target mean/std bugs

configs/kitti/dv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py

@@ -128,7 +128,7 @@ train_pipeline = [
     dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
     dict(type='PointShuffle'),
     dict(type='DefaultFormatBundle3D', class_names=class_names),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes', 'gt_labels']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d']),
 ]
 test_pipeline = [
     dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
@@ -136,7 +136,7 @@ test_pipeline = [
         type='DefaultFormatBundle3D',
         class_names=class_names,
         with_label=False),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d']),
 ]
 
 data = dict(
@@ -177,13 +177,13 @@ optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
 # learning policy
 lr_config = dict(
     policy='cyclic',
-    target_ratio=[10, 1e-4],
+    target_ratio=(10, 1e-4),
     cyclic_times=1,
     step_ratio_up=0.4,
 )
 momentum_config = dict(
     policy='cyclic',
-    target_ratio=[0.85 / 0.95, 1],
+    target_ratio=(0.85 / 0.95, 1),
     cyclic_times=1,
     step_ratio_up=0.4,
 )

configs/kitti/dv_second_secfpn_6x8_80e_kitti-3d-car.py

@@ -94,7 +94,7 @@ input_modality = dict(
     use_lidar=True,
     use_depth=False,
     use_lidar_intensity=True,
-    use_camera=False,
+    use_camera=True,
 )
 db_sampler = dict(
     root_path=data_root,
@@ -125,7 +125,7 @@ train_pipeline = [
     dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
     dict(type='PointShuffle'),
     dict(type='DefaultFormatBundle3D', class_names=class_names),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes', 'gt_labels']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d']),
 ]
 test_pipeline = [
     dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
@@ -133,7 +133,7 @@ test_pipeline = [
         type='DefaultFormatBundle3D',
         class_names=class_names,
         with_label=False),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d']),
 ]
 
 data = dict(
@@ -173,13 +173,13 @@ optimizer = dict(type='AdamW', lr=lr, betas=(0.95, 0.99), weight_decay=0.01)
 optimizer_config = dict(grad_clip=dict(max_norm=10, norm_type=2))
 lr_config = dict(
     policy='cyclic',
-    target_ratio=[10, 1e-4],
+    target_ratio=(10, 1e-4),
     cyclic_times=1,
     step_ratio_up=0.4,
 )
 momentum_config = dict(
     policy='cyclic',
-    target_ratio=[0.85 / 0.95, 1],
+    target_ratio=(0.85 / 0.95, 1),
     cyclic_times=1,
     step_ratio_up=0.4,
 )

configs/kitti/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py

@@ -125,7 +125,7 @@ train_pipeline = [
     dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
     dict(type='PointShuffle'),
     dict(type='DefaultFormatBundle3D', class_names=class_names),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes', 'gt_labels']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d']),
 ]
 test_pipeline = [
     dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
@@ -133,7 +133,7 @@ test_pipeline = [
         type='DefaultFormatBundle3D',
         class_names=class_names,
         with_label=False),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d']),
 ]
 
 data = dict(
@@ -178,13 +178,13 @@ optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
 # learning policy
 lr_config = dict(
     policy='cyclic',
-    target_ratio=[10, 1e-4],
+    target_ratio=(10, 1e-4),
     cyclic_times=1,
     step_ratio_up=0.4,
 )
 momentum_config = dict(
     policy='cyclic',
-    target_ratio=[0.85 / 0.95, 1],
+    target_ratio=(0.85 / 0.95, 1),
     cyclic_times=1,
     step_ratio_up=0.4,
 )

configs/kitti/hv_second_secfpn_6x8_80e_kitti-3d-car.py

@@ -123,7 +123,7 @@ train_pipeline = [
     dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
     dict(type='PointShuffle'),
     dict(type='DefaultFormatBundle3D', class_names=class_names),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes', 'gt_labels']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d']),
 ]
 test_pipeline = [
     dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
@@ -131,7 +131,7 @@ test_pipeline = [
         type='DefaultFormatBundle3D',
         class_names=class_names,
         with_label=False),
-    dict(type='Collect3D', keys=['points', 'gt_bboxes']),
+    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d']),
 ]
 
 data = dict(
@@ -171,13 +171,13 @@ optimizer = dict(type='AdamW', lr=lr, betas=(0.95, 0.99), weight_decay=0.01)
 optimizer_config = dict(grad_clip=dict(max_norm=10, norm_type=2))
 lr_config = dict(
     policy='cyclic',
-    target_ratio=[10, 1e-4],
+    target_ratio=(10, 1e-4),
     cyclic_times=1,
     step_ratio_up=0.4,
 )
 momentum_config = dict(
     policy='cyclic',
-    target_ratio=[0.85 / 0.95, 1],
+    target_ratio=(0.85 / 0.95, 1),
     cyclic_times=1,
     step_ratio_up=0.4,
 )

mmdet3d/models/anchor_heads/boxvelo_head.py

@@ -14,12 +14,6 @@ class Anchor3DVeloHead(SECONDHead):
     Args:
         in_channels (int): Number of channels in the input feature map.
         feat_channels (int): Number of channels of the feature map.
-        anchor_scales (Iterable): Anchor scales.
-        anchor_ratios (Iterable): Anchor aspect ratios.
-        anchor_strides (Iterable): Anchor strides.
-        anchor_base_sizes (Iterable): Anchor base sizes.
-        target_means (Iterable): Mean values of regression targets.
-        target_stds (Iterable): Std values of regression targets.
         loss_cls (dict): Config of classification loss.
         loss_bbox (dict): Config of localization loss.
     """  # noqa: W605
@@ -127,9 +121,7 @@ class Anchor3DVeloHead(SECONDHead):
                 scores = scores[topk_inds, :]
                 dir_cls_score = dir_cls_score[topk_inds]
 
-            bboxes = self.bbox_coder.decode(anchors, bbox_pred,
-                                            self.target_means,
-                                            self.target_stds)
+            bboxes = self.bbox_coder.decode(anchors, bbox_pred)
             mlvl_bboxes.append(bboxes)
             mlvl_scores.append(scores)
             mlvl_dir_scores.append(dir_cls_score)

mmdet3d/models/anchor_heads/second_head.py

@@ -19,12 +19,6 @@ class SECONDHead(nn.Module, AnchorTrainMixin):
     Args:
         in_channels (int): Number of channels in the input feature map.
         feat_channels (int): Number of channels of the feature map.
-        anchor_scales (Iterable): Anchor scales.
-        anchor_ratios (Iterable): Anchor aspect ratios.
-        anchor_strides (Iterable): Anchor strides.
-        anchor_base_sizes (Iterable): Anchor base sizes.
-        target_means (Iterable): Mean values of regression targets.
-        target_stds (Iterable): Std values of regression targets.
         loss_cls (dict): Config of classification loss.
         loss_bbox (dict): Config of localization loss.
     """  # noqa: W605
@@ -216,8 +210,6 @@ class SECONDHead(nn.Module, AnchorTrainMixin):
             anchor_list,
             gt_bboxes,
             input_metas,
-            self.target_means,
-            self.target_stds,
             gt_bboxes_ignore_list=gt_bboxes_ignore,
             gt_labels_list=gt_labels,
             num_classes=self.num_classes,

mmdet3d/models/anchor_heads/train_mixins.py

@@ -11,8 +11,6 @@ class AnchorTrainMixin(object):
                          anchor_list,
                          gt_bboxes_list,
                          input_metas,
-                         target_means,
-                         target_stds,
                          gt_bboxes_ignore_list=None,
                          gt_labels_list=None,
                          label_channels=1,
@@ -24,8 +22,6 @@ class AnchorTrainMixin(object):
             anchor_list (list[list]): Multi level anchors of each image.
             gt_bboxes_list (list[Tensor]): Ground truth bboxes of each image.
             img_metas (list[dict]): Meta info of each image.
-            target_means (Iterable): Mean value of regression targets.
-            target_stds (Iterable): Std value of regression targets.
 
         Returns:
             tuple
@@ -57,8 +53,6 @@ class AnchorTrainMixin(object):
              gt_bboxes_ignore_list,
              gt_labels_list,
              input_metas,
-             target_means=target_means,
-             target_stds=target_stds,
              label_channels=label_channels,
              num_classes=num_classes,
              sampling=sampling)
@@ -89,8 +83,6 @@ class AnchorTrainMixin(object):
                                 gt_bboxes_ignore,
                                 gt_labels,
                                 input_meta,
-                                target_means,
-                                target_stds,
                                 label_channels=1,
                                 num_classes=1,
                                 sampling=True):
@@ -111,13 +103,12 @@ class AnchorTrainMixin(object):
                     anchor_targets = self.anchor_target_single_assigner(
                         assigner, current_anchors, gt_bboxes[gt_per_cls, :],
                         gt_bboxes_ignore, gt_labels[gt_per_cls], input_meta,
-                        target_means, target_stds, label_channels, num_classes,
-                        sampling)
+                        label_channels, num_classes, sampling)
                 else:
                     anchor_targets = self.anchor_target_single_assigner(
                         assigner, current_anchors, gt_bboxes, gt_bboxes_ignore,
-                        gt_labels, input_meta, target_means, target_stds,
-                        label_channels, num_classes, sampling)
+                        gt_labels, input_meta, label_channels, num_classes,
+                        sampling)
 
                 (labels, label_weights, bbox_targets, bbox_weights,
                  dir_targets, dir_weights, pos_inds, neg_inds) = anchor_targets
@@ -156,8 +147,7 @@ class AnchorTrainMixin(object):
         else:
             return self.anchor_target_single_assigner(
                 self.bbox_assigner, anchors, gt_bboxes, gt_bboxes_ignore,
-                gt_labels, input_meta, target_means, target_stds,
-                label_channels, num_classes, sampling)
+                gt_labels, input_meta, label_channels, num_classes, sampling)
 
     def anchor_target_single_assigner(self,
                                       bbox_assigner,
@@ -166,8 +156,6 @@ class AnchorTrainMixin(object):
                                       gt_bboxes_ignore,
                                       gt_labels,
                                       input_meta,
-                                      target_means,
-                                      target_stds,
                                       label_channels=1,
                                       num_classes=1,
                                       sampling=True):

tools/slurm_train.sh

 #!/usr/bin/env bash
 
 set -x
-export PYTHONPATH=`pwd`:$PYTHONPATH
 
 PARTITION=$1
 JOB_NAME=$2
@@ -20,4 +19,4 @@ srun -p ${PARTITION} \
     --ntasks-per-node=${GPUS_PER_NODE} \
     --kill-on-bad-exit=1 \
     ${SRUN_ARGS} \
-    python -u tools/train.py ${CONFIG} --work_dir=${WORK_DIR} --launcher="slurm" ${PY_ARGS}
+    python -u tools/train.py ${CONFIG} --work-dir=${WORK_DIR} --launcher="slurm" ${PY_ARGS}