Skip to content

GitLab

"...visualglm6b_pytorch.git" did not exist on "3fe98be6a62cf552f467d74b252ed1d5c076090a"
Commit 4040dbda authored by zhangwenwei's avatar zhangwenwei
Browse files

Refactor anchor generator and box coder

parent 148fea12
Hide whitespace changes
Inline Side-by-side
Showing with 291 additions and 214 deletions
configs/kitti/dv_mvx-v2_second_secfpn_fpn-fusion_adamw_2x8_80e_kitti-3d-3class.py
View file @ 4040dbda
......@@ -72,18 +72,21 @@ model = dict(
feat_channels=512,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -1.78, 70.4, 40.0, -1.78],
],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -1.78, 70.4, 40.0, -1.78],
],
strides=[2],
sizes=[[0.6, 0.8, 1.73], [0.6, 1.76, 1.73], [1.6, 3.9, 1.56]],
rotations=[0, 1.57],
reshape_out=False),
assigner_per_size=True,
anchor_strides=[2],
anchor_sizes=[[0.6, 0.8, 1.73], [0.6, 1.76, 1.73], [1.6, 3.9, 1.56]],
anchor_rotations=[0, 1.57],
diff_rad_by_sin=True,
assign_per_class=True,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder', ),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......
configs/kitti/dv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py
View file @ 4040dbda
......@@ -43,12 +43,15 @@ model = dict(
feat_channels=384,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[0, -39.68, -1.78, 69.12, 39.68, -1.78],
anchor_strides=[2],
anchor_sizes=[[1.6, 3.9, 1.56]],
anchor_rotations=[0, 1.57],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[[0, -39.68, -1.78, 69.12, 39.68, -1.78]],
strides=[2],
sizes=[[1.6, 3.9, 1.56]],
rotations=[0, 1.57],
reshape_out=True),
diff_rad_by_sin=True,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder', ),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......@@ -116,7 +119,7 @@ train_pipeline = [
loc_noise_std=[0.25, 0.25, 0.25],
global_rot_range=[0.0, 0.0],
rot_uniform_noise=[-0.15707963267, 0.15707963267]),
dict(type='PointsRandomFlip', flip_ratio=0.5),
dict(type='RandomFlip3D', flip_ratio=0.5),
dict(
type='GlobalRotScale',
rot_uniform_noise=[-0.78539816, 0.78539816],
......
configs/kitti/dv_second_secfpn_2x8_cosine_80e_kitti-3d-3class.py
View file @ 4040dbda
......@@ -41,18 +41,21 @@ model = dict(
feat_channels=512,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -1.78, 70.4, 40.0, -1.78],
],
anchor_strides=[2],
anchor_sizes=[[0.6, 0.8, 1.73], [0.6, 1.76, 1.73], [1.6, 3.9, 1.56]],
anchor_rotations=[0, 1.57],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -0.6, 70.4, 40.0, -0.6],
[0, -40.0, -1.78, 70.4, 40.0, -1.78],
],
strides=[2],
sizes=[[0.6, 0.8, 1.73], [0.6, 1.76, 1.73], [1.6, 3.9, 1.56]],
rotations=[0, 1.57],
reshape_out=False),
diff_rad_by_sin=True,
assigner_per_size=True,
assign_per_class=True,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder', ),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......
configs/kitti/dv_second_secfpn_6x8_80e_kitti-3d-car.py
View file @ 4040dbda
......@@ -41,12 +41,15 @@ model = dict(
feat_channels=512,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[0, -40.0, -1.78, 70.4, 40.0, -1.78],
anchor_strides=[2],
anchor_sizes=[[1.6, 3.9, 1.56]],
anchor_rotations=[0, 1.57],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[[0, -40.0, -1.78, 70.4, 40.0, -1.78]],
strides=[2],
sizes=[[1.6, 3.9, 1.56]],
rotations=[0, 1.57],
reshape_out=True),
diff_rad_by_sin=True,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......@@ -113,7 +116,7 @@ train_pipeline = [
loc_noise_std=[1.0, 1.0, 0.5],
global_rot_range=[0.0, 0.0],
rot_uniform_noise=[-0.78539816, 0.78539816]),
dict(type='PointsRandomFlip', flip_ratio=0.5),
dict(type='RandomFlip3D', flip_ratio=0.5),
dict(
type='GlobalRotScale',
rot_uniform_noise=[-0.78539816, 0.78539816],
......
configs/kitti/faster_rcnn_r50_fpn_caffe_1x_kitti-2d-3class_coco-3x-pretrain.py
View file @ 4040dbda
......@@ -2,8 +2,7 @@
norm_cfg = dict(type='BN', requires_grad=False)
model = dict(
type='FasterRCNN',
pretrained=('./pretrain_detectron/'
'ImageNetPretrained/MSRA/resnet50_msra.pth'),
pretrained=('open-mmlab://resnet50_caffe_bgr'),
backbone=dict(
type='ResNet',
depth=50,
......@@ -22,11 +21,15 @@ model = dict(
type='RPNHead',
in_channels=256,
feat_channels=256,
anchor_scales=[8],
anchor_ratios=[1 / 3, 0.5, 1.0, 2.0, 3.0],
anchor_strides=[4, 8, 16, 32, 64],
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0],
anchor_generator=dict(
type='AnchorGenerator',
scales=[8],
ratios=[0.5, 1.0, 2.0],
strides=[4, 8, 16, 32, 64]),
bbox_coder=dict(
type='DeltaXYWHBBoxCoder',
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0]),
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
loss_bbox=dict(type='L1Loss', loss_weight=1.0)),
......@@ -43,8 +46,10 @@ model = dict(
fc_out_channels=1024,
roi_feat_size=7,
num_classes=80,
target_means=[0., 0., 0., 0.],
target_stds=[0.1, 0.1, 0.2, 0.2],
bbox_coder=dict(
type='DeltaXYWHBBoxCoder',
target_means=[0., 0., 0., 0.],
target_stds=[0.1, 0.1, 0.2, 0.2]),
reg_class_agnostic=False,
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
......
configs/kitti/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py
View file @ 4040dbda
......@@ -42,12 +42,15 @@ model = dict(
feat_channels=384,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[0, -39.68, -1.78, 69.12, 39.68, -1.78],
anchor_strides=[2],
anchor_sizes=[[1.6, 3.9, 1.56]],
anchor_rotations=[0, 1.57],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[[0, -39.68, -1.78, 69.12, 39.68, -1.78]],
strides=[2],
sizes=[[1.6, 3.9, 1.56]],
rotations=[0, 1.57],
reshape_out=True),
diff_rad_by_sin=True,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......@@ -113,7 +116,7 @@ train_pipeline = [
loc_noise_std=[0.25, 0.25, 0.25],
global_rot_range=[0.0, 0.0],
rot_uniform_noise=[-0.15707963267, 0.15707963267]),
dict(type='PointsRandomFlip', flip_ratio=0.5),
dict(type='RandomFlip3D', flip_ratio=0.5),
dict(
type='GlobalRotScale',
rot_uniform_noise=[-0.78539816, 0.78539816],
......
configs/kitti/hv_second_secfpn_6x8_80e_kitti-3d-car.py
View file @ 4040dbda
......@@ -41,12 +41,15 @@ model = dict(
feat_channels=512,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[0, -40.0, -1.78, 70.4, 40.0, -1.78],
anchor_strides=[2],
anchor_sizes=[[1.6, 3.9, 1.56]],
anchor_rotations=[0, 1.57],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[[0, -40.0, -1.78, 70.4, 40.0, -1.78]],
strides=[2],
sizes=[[1.6, 3.9, 1.56]],
rotations=[0, 1.57],
reshape_out=True),
diff_rad_by_sin=True,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......@@ -111,7 +114,7 @@ train_pipeline = [
loc_noise_std=[1.0, 1.0, 0.5],
global_rot_range=[0.0, 0.0],
rot_uniform_noise=[-0.78539816, 0.78539816]),
dict(type='PointsRandomFlip', flip_ratio=0.5),
dict(type='RandomFlip3D', flip_ratio=0.5),
dict(
type='GlobalRotScale',
rot_uniform_noise=[-0.78539816, 0.78539816],
......
configs/nus/faster_rcnn_r50_fpn_caffe_2x8_1x_nus.py
View file @ 4040dbda
......@@ -2,8 +2,7 @@
norm_cfg = dict(type='BN', requires_grad=False)
model = dict(
type='FasterRCNN',
pretrained=('./pretrain_detectron/'
'ImageNetPretrained/MSRA/resnet50_msra.pth'),
pretrained=('open-mmlab://resnet50_caffe_bgr'),
backbone=dict(
type='ResNet',
depth=50,
......@@ -22,11 +21,15 @@ model = dict(
type='RPNHead',
in_channels=256,
feat_channels=256,
anchor_scales=[8],
anchor_ratios=[0.5, 1.0, 2.0],
anchor_strides=[4, 8, 16, 32, 64],
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0],
anchor_generator=dict(
type='AnchorGenerator',
scales=[8],
ratios=[0.5, 1.0, 2.0],
strides=[4, 8, 16, 32, 64]),
bbox_coder=dict(
type='DeltaXYWHBBoxCoder',
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0]),
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0),
loss_bbox=dict(type='L1Loss', loss_weight=1.0)),
......@@ -42,9 +45,11 @@ model = dict(
in_channels=256,
fc_out_channels=1024,
roi_feat_size=7,
num_classes=80,
target_means=[0., 0., 0., 0.],
target_stds=[0.1, 0.1, 0.2, 0.2],
num_classes=10,
bbox_coder=dict(
type='DeltaXYWHBBoxCoder',
target_means=[0., 0., 0., 0.],
target_stds=[0.1, 0.1, 0.2, 0.2]),
reg_class_agnostic=False,
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
......@@ -105,12 +110,14 @@ test_cfg = dict(
# e.g., nms=dict(type='soft_nms', iou_thr=0.5, min_score=0.05)
)
# dataset settings
dataset_type = 'NuScenes2DDataset'
dataset_type = 'CocoDataset'
data_root = 'data/nuscenes/'
# Values to be used for image normalization (BGR order)
# Default mean pixel values are from ImageNet: [103.53, 116.28, 123.675]
# When using pre-trained models in Detectron1 or any MSRA models,
# std has been absorbed into its conv1 weights, so the std needs to be set 1.
classes = ('car', 'truck', 'trailer', 'bus', 'construction_vehicle', 'bicycle',
'motorcycle', 'pedestrian', 'traffic_cone', 'barrier')
img_norm_cfg = dict(
mean=[103.530, 116.280, 123.675], std=[1.0, 1.0, 1.0], to_rgb=False)
train_pipeline = [
......@@ -147,14 +154,17 @@ data = dict(
workers_per_gpu=2,
train=dict(
type=dataset_type,
classes=classes,
ann_file=data_root + 'nuscenes_infos_train.coco.json',
pipeline=train_pipeline),
val=dict(
type=dataset_type,
classes=classes,
ann_file=data_root + 'nuscenes_infos_val.coco.json',
pipeline=test_pipeline),
test=dict(
type=dataset_type,
classes=classes,
ann_file=data_root + 'nuscenes_infos_val.coco.json',
pipeline=test_pipeline))
# optimizer
......
configs/nus/hv_pointpillars_secfpn_sbn-all_4x8_20e_nus-3d.py
View file @ 4040dbda
......@@ -51,33 +51,35 @@ model = dict(
feat_channels=384,
use_direction_classifier=True,
encode_bg_as_zeros=True,
anchor_range=[
[-49.6, -49.6, -1.80032795, 49.6, 49.6, -1.80032795], # car
[-49.6, -49.6, -1.74440365, 49.6, 49.6, -1.74440365], # truck
[-49.6, -49.6, -1.68526504, 49.6, 49.6, -1.68526504], # trailer
[-49.6, -49.6, -1.67339111, 49.6, 49.6, -1.67339111], # bicycle
[-49.6, -49.6, -1.61785072, 49.6, 49.6, -1.61785072], # pedestrian
[-49.6, -49.6, -1.80984986, 49.6, 49.6,
-1.80984986], # traffic_cone
[-49.6, -49.6, -1.763965, 49.6, 49.6, -1.763965], # barrier
],
anchor_strides=[2],
anchor_sizes=[
[1.95017717, 4.60718145, 1.72270761], # car
[2.4560939, 6.73778078, 2.73004906], # truck
[2.87427237, 12.01320693, 3.81509561], # trailer
[0.60058911, 1.68452161, 1.27192197], # bicycle
[0.66344886, 0.7256437, 1.75748069], # pedestrian
[0.39694519, 0.40359262, 1.06232151], # traffic_cone
[2.49008838, 0.48578221, 0.98297065], # barrier
],
anchor_custom_values=[0, 0],
anchor_rotations=[0, 1.57],
anchor_generator=dict(
type='Anchor3DRangeGenerator',
ranges=[
[-49.6, -49.6, -1.80032795, 49.6, 49.6, -1.80032795],
[-49.6, -49.6, -1.74440365, 49.6, 49.6, -1.74440365],
[-49.6, -49.6, -1.68526504, 49.6, 49.6, -1.68526504],
[-49.6, -49.6, -1.67339111, 49.6, 49.6, -1.67339111],
[-49.6, -49.6, -1.61785072, 49.6, 49.6, -1.61785072],
[-49.6, -49.6, -1.80984986, 49.6, 49.6, -1.80984986],
[-49.6, -49.6, -1.763965, 49.6, 49.6, -1.763965],
],
strides=[2],
sizes=[
[1.95017717, 4.60718145, 1.72270761], # car
[2.4560939, 6.73778078, 2.73004906], # truck
[2.87427237, 12.01320693, 3.81509561], # trailer
[0.60058911, 1.68452161, 1.27192197], # bicycle
[0.66344886, 0.7256437, 1.75748069], # pedestrian
[0.39694519, 0.40359262, 1.06232151], # traffic_cone
[2.49008838, 0.48578221, 0.98297065], # barrier
],
custom_values=[0, 0],
rotations=[0, 1.57],
reshape_out=True),
assigner_per_size=False,
diff_rad_by_sin=True,
dir_offset=0.7854, # pi/4
dir_limit_offset=0,
bbox_coder=dict(type='Residual3DBoxCoder', ),
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder', ),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......
configs/nus/retinanet_r50_fpn_caffe_2x8_1x_nus.py
View file @ 4040dbda
......@@ -2,8 +2,7 @@
norm_cfg = dict(type='BN', requires_grad=False)
model = dict(
type='RetinaNet',
pretrained=('./pretrain_detectron/'
'ImageNetPretrained/MSRA/resnet50_msra.pth'),
pretrained=('open-mmlab://resnet50_caffe_bgr'),
backbone=dict(
type='ResNet',
depth=50,
......@@ -26,12 +25,16 @@ model = dict(
in_channels=256,
stacked_convs=4,
feat_channels=256,
octave_base_scale=4,
scales_per_octave=3,
anchor_ratios=[0.5, 1.0, 2.0],
anchor_strides=[8, 16, 32, 64, 128],
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0],
anchor_generator=dict(
type='AnchorGenerator',
octave_base_scale=4,
scales_per_octave=3,
ratios=[0.5, 1.0, 2.0],
strides=[8, 16, 32, 64, 128]),
bbox_coder=dict(
type='DeltaXYWHBBoxCoder',
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0]),
loss_cls=dict(
type='FocalLoss',
use_sigmoid=True,
......
mmdet3d/apis/__init__.py
View file @ 4040dbda
from .train import train_detector
from .train import batch_processor, train_detector
__all__ = [
'train_detector',
]
__all__ = ['batch_processor', 'train_detector']
mmdet3d/core/anchor/__init__.py
View file @ 4040dbda
from .anchor_3d_generator import (AlignedAnchorGeneratorRange,
AnchorGeneratorRange)
from mmdet.core.anchor import build_anchor_generator
from .anchor_3d_generator import (AlignedAnchor3DRangeGenerator,
Anchor3DRangeGenerator)
__all__ = [
'AlignedAnchorGeneratorRange', 'AnchorGeneratorRange',
'AlignedAnchor3DRangeGenerator', 'Anchor3DRangeGenerator',
'build_anchor_generator'
]
def build_anchor_generator(cfg, **kwargs):
from . import anchor_3d_generator
import mmcv
if isinstance(cfg, dict):
return mmcv.runner.obj_from_dict(
cfg, anchor_3d_generator, default_args=kwargs)
else:
raise TypeError('Invalid type {} for building a sampler'.format(
type(cfg)))
mmdet3d/core/anchor/anchor_3d_generator.py
View file @ 4040dbda
import mmcv
import torch
from mmdet.core.anchor import ANCHOR_GENERATORS
class AnchorGeneratorRange(object):
@ANCHOR_GENERATORS.register_module
class Anchor3DRangeGenerator(object):
"""3D Anchor Generator by range
This anchor generator generates anchors by the given range in different
feature levels.
Due the convention in 3D detection, different anchor sizes are related to
different ranges for different categories. However we find this setting
does not effect the performance much in some datasets, e.g., nuScenes.
Args:
ranges (list[list[float]]): Ranges of different anchors.
The ranges are the same across different feature levels. But may
vary for different anchor sizes if size_per_range is True.
sizes (list[list[float]]): 3D sizes of anchors.
strides (list[int]): Strides of anchors in different feature levels.
rotations (list(float)): Rotations of anchors in a feature grid.
custom_values (tuple(float)): Customized values of that anchor. For
example, in nuScenes the anchors have velocities.
reshape_out (bool): Whether to reshape the output into (Nx4)
size_per_range: Whether to use separate ranges for different sizes.
If size_per_range is True, the ranges should have the same length
as the sizes, if not, it will be duplicated.
"""
def __init__(self,
anchor_ranges,
sizes=((1.6, 3.9, 1.56), ),
stride=2,
rotations=(0, 3.1415926 / 2),
ranges,
sizes=[[1.6, 3.9, 1.56]],
strides=[2],
rotations=[0, 1.5707963],
custom_values=(),
cache_anchor=False):
reshape_out=True,
size_per_range=True):
assert mmcv.is_list_of(ranges, list)
if size_per_range:
if len(sizes) != len(ranges):
assert len(ranges) == 1
ranges = ranges * len(sizes)
assert len(ranges) == len(sizes)
else:
assert len(ranges) == 1
assert mmcv.is_list_of(sizes, list)
assert isinstance(strides, list)
self.sizes = sizes
self.stride = stride
self.anchor_ranges = anchor_ranges
if len(anchor_ranges) != len(sizes):
self.anchor_ranges = anchor_ranges * len(sizes)
self.strides = strides
self.ranges = ranges
self.rotations = rotations
self.custom_values = custom_values
self.cache_anchor = cache_anchor
self.cached_anchors = None
self.reshape_out = reshape_out
self.size_per_range = size_per_range
def __repr__(self):
s = self.__class__.__name__ + '('
s += 'anchor_range={}, '.format(self.anchor_ranges)
s += 'stride={}, '.format(self.stride)
s += 'sizes={}, '.format(self.sizes)
s += 'rotations={})'.format(self.rotations)
s += f'anchor_range={self.ranges},\n'
s += f'strides={self.strides},\n'
s += f'sizes={self.sizes},\n'
s += f'rotations={self.rotations},\n'
s += f'reshape_out={self.reshape_out},\n'
s += f'size_per_range={self.size_per_range})'
return s
@property
......@@ -34,40 +73,68 @@ class AnchorGeneratorRange(object):
num_size = torch.tensor(self.sizes).reshape(-1, 3).size(0)
return num_rot * num_size
def grid_anchors(self, feature_map_size, device='cuda'):
@property
def num_levels(self):
return len(self.strides)
def grid_anchors(self, featmap_sizes, device='cuda'):
"""Generate grid anchors in multiple feature levels
Args:
featmap_sizes (list[tuple]): List of feature map sizes in
multiple feature levels.
device (str): Device where the anchors will be put on.
Return:
list[torch.Tensor]: Anchors in multiple feature levels.
The sizes of each tensor should be [N, 4], where
N = width * height * num_base_anchors, width and height
are the sizes of the corresponding feature lavel,
num_base_anchors is the number of anchors for that level.
"""
assert self.num_levels == len(featmap_sizes)
multi_level_anchors = []
for i in range(self.num_levels):
anchors = self.single_level_grid_anchors(
featmap_sizes[i], self.strides[i], device=device)
if self.reshape_out:
anchors = anchors.reshape(-1, anchors.size(-1))
multi_level_anchors.append(anchors)
return multi_level_anchors
def single_level_grid_anchors(self, featmap_size, stride, device='cuda'):
# We reimplement the anchor generator using torch in cuda
# torch: 0.6975 s for 1000 times
# numpy: 4.3345 s for 1000 times
# which is ~5 times faster than numpy implementation
if (self.cache_anchor and self.cached_anchors):
return self.cached_anchors
if not isinstance(self.anchor_ranges[0], list):
# which is ~5 times faster than the numpy implementation
if not self.size_per_range:
return self.anchors_single_range(
feature_map_size,
self.anchor_ranges,
featmap_size,
self.ranges[0],
stride,
self.sizes,
self.rotations,
device=device)
assert len(self.sizes) == len(self.anchor_ranges)
mr_anchors = []
for anchor_range, anchor_size in zip(self.anchor_ranges, self.sizes):
for anchor_range, anchor_size in zip(self.ranges, self.sizes):
mr_anchors.append(
self.anchors_single_range(
feature_map_size,
featmap_size,
anchor_range,
stride,
anchor_size,
self.rotations,
device=device))
mr_anchors = torch.cat(mr_anchors, dim=-3)
if self.cache_anchor and not self.cached_anchors:
self.cached_anchors = mr_anchors
return mr_anchors
def anchors_single_range(self,
feature_size,
anchor_range,
sizes=((1.6, 3.9, 1.56), ),
rotations=(0, 3.1415927 / 2),
stride=1,
sizes=[[1.6, 3.9, 1.56]],
rotations=[0, 1.5707963],
device='cuda'):
"""Generate anchors in a single range
Args:
......@@ -106,7 +173,6 @@ class AnchorGeneratorRange(object):
ret = torch.cat(rets, dim=-1).permute([2, 1, 0, 3, 4, 5])
# [1, 200, 176, N, 2, 7] for kitti after permute
# ret = ret.reshape(-1, 7)
if len(self.custom_values) > 0:
custom_ndim = len(self.custom_values)
......@@ -117,17 +183,42 @@ class AnchorGeneratorRange(object):
return ret
class AlignedAnchorGeneratorRange(AnchorGeneratorRange):
@ANCHOR_GENERATORS.register_module
class AlignedAnchor3DRangeGenerator(Anchor3DRangeGenerator):
"""Aligned 3D Anchor Generator by range
This anchor generator uses a different manner to generate the positions
of anchors' centers from `Anchor3DRangeGenerator`.
Note:
The `align` means that the anchor's center is aligned with the voxel grid,
which is also the feature grid. The previous implementation of
`Anchor3DRangeGenerator` do not generate the anchors' center according
to the voxel grid. Rather, it generates the center by uniformly
distributing the anchors inside the minimum and maximum anchor ranges
according to the feature map sizes.
However, this makes the anchors center does not match the feature grid.
The AlignedAnchor3DRangeGenerator add + 1 when using the feature map sizes
to obtain the corners of the voxel grid. Then it shift the coordinates to
the center of voxel grid of use the left up corner to distribute anchors.
Args:
anchor_corner (bool): Whether to align with the corner of the voxel
grid. By default it is False and the anchor's center will be
the same as the corresponding voxel's center, which is also the
center of the corresponding greature grid.
"""
def __init__(self, shift_center=True, **kwargs):
super(AlignedAnchorGeneratorRange, self).__init__(**kwargs)
self.shift_center = shift_center
def __init__(self, align_corner=False, **kwargs):
super(AlignedAnchor3DRangeGenerator, self).__init__(**kwargs)
self.align_corner = align_corner
def anchors_single_range(self,
feature_size,
anchor_range,
sizes=((1.6, 3.9, 1.56), ),
rotations=(0, 3.1415927 / 2),
stride,
sizes=[[1.6, 3.9, 1.56]],
rotations=[0, 1.5707963],
device='cuda'):
"""Generate anchors in a single range
Args:
......@@ -155,11 +246,11 @@ class AlignedAnchorGeneratorRange(AnchorGeneratorRange):
anchor_range[3],
feature_size[2] + 1,
device=device)
sizes = torch.tensor(sizes, device=device).reshape(-1, 3) * self.stride
sizes = torch.tensor(sizes, device=device).reshape(-1, 3) * stride
rotations = torch.tensor(rotations, device=device)
# shift the anchor center
if self.shift_center:
if not self.align_corner:
z_shift = (z_centers[1] - z_centers[0]) / 2
y_shift = (y_centers[1] - y_centers[0]) / 2
x_shift = (x_centers[1] - x_centers[0]) / 2
......@@ -187,7 +278,6 @@ class AlignedAnchorGeneratorRange(AnchorGeneratorRange):
ret = torch.cat(rets, dim=-1).permute([2, 1, 0, 3, 4, 5])
# [1, 200, 176, N, 2, 7] for kitti after permute
# ret = ret.reshape(-1, 7)
if len(self.custom_values) > 0:
custom_ndim = len(self.custom_values)
......
mmdet3d/core/bbox/__init__.py
View file @ 4040dbda
from . import box_torch_ops
from .assigners import AssignResult, BaseAssigner, MaxIoUAssigner
from .coders import Residual3DBoxCoder
from .coders import DeltaXYZWLHRBBoxCoder
# from .bbox_target import bbox_target
from .iou_calculators import (BboxOverlaps3D, BboxOverlapsNearest3D,
bbox_overlaps_3d, bbox_overlaps_nearest_3d)
......@@ -18,7 +18,7 @@ __all__ = [
'PseudoSampler', 'RandomSampler', 'InstanceBalancedPosSampler',
'IoUBalancedNegSampler', 'CombinedSampler', 'SamplingResult',
'build_assigner', 'build_sampler', 'assign_and_sample', 'box_torch_ops',
'build_bbox_coder', 'Residual3DBoxCoder', 'boxes3d_to_bev_torch_lidar',
'build_bbox_coder', 'DeltaXYZWLHRBBoxCoder', 'boxes3d_to_bev_torch_lidar',
'BboxOverlapsNearest3D', 'BboxOverlaps3D', 'bbox_overlaps_nearest_3d',
'bbox_overlaps_3d'
]
mmdet3d/core/bbox/assign_sampling.py
View file @ 4040dbda
......@@ -14,7 +14,7 @@ def build_assigner(cfg, **kwargs):
def build_bbox_coder(cfg, **kwargs):
if isinstance(cfg, coders.Residual3DBoxCoder):
if isinstance(cfg, coders.DeltaXYZWLHRBBoxCoder):
return cfg
elif isinstance(cfg, dict):
return mmcv.runner.obj_from_dict(cfg, coders, default_args=kwargs)
......
mmdet3d/core/bbox/coders/__init__.py
View file @ 4040dbda
from .box_coder import Residual3DBoxCoder
from mmdet.core.bbox import build_bbox_coder
from .delta_xywh_bbox_coder import DeltaXYZWLHRBBoxCoder
__all__ = ['Residual3DBoxCoder']
__all__ = ['build_bbox_coder', 'DeltaXYZWLHRBBoxCoder']
mmdet3d/core/bbox/coders/box_coder.py → mmdet3d/core/bbox/coders/delta_xywh_bbox_coder.py
View file @ 4040dbda
import numpy as np
import torch
from mmdet.core.bbox import BaseBBoxCoder
from mmdet.core.bbox.builder import BBOX_CODERS
class Residual3DBoxCoder(object):
def __init__(self, code_size=7, mean=None, std=None):
super().__init__()
self.code_size = code_size
self.mean = mean
self.std = std
@staticmethod
def encode_np(boxes, anchors):
"""
:param boxes: (N, 7) x, y, z, w, l, h, r
:param anchors: (N, 7)
:return:
"""
# need to convert boxes to z-center format
xa, ya, za, wa, la, ha, ra = np.split(anchors, 7, axis=-1)
xg, yg, zg, wg, lg, hg, rg = np.split(boxes, 7, axis=-1)
zg = zg + hg / 2
za = za + ha / 2
diagonal = np.sqrt(la**2 + wa**2) # 4.3
xt = (xg - xa) / diagonal
yt = (yg - ya) / diagonal
zt = (zg - za) / ha # 1.6
lt = np.log(lg / la)
wt = np.log(wg / wa)
ht = np.log(hg / ha)
rt = rg - ra
return np.concatenate([xt, yt, zt, wt, lt, ht, rt], axis=-1)
@staticmethod
def decode_np(box_encodings, anchors):
"""
:param box_encodings: (N, 7) x, y, z, w, l, h, r
:param anchors: (N, 7)
:return:
"""
# need to convert box_encodings to z-bottom format
xa, ya, za, wa, la, ha, ra = np.split(anchors, 7, axis=-1)
xt, yt, zt, wt, lt, ht, rt = np.split(box_encodings, 7, axis=-1)
@BBOX_CODERS.register_module
class DeltaXYZWLHRBBoxCoder(BaseBBoxCoder):
za = za + ha / 2
diagonal = np.sqrt(la**2 + wa**2)
xg = xt * diagonal + xa
yg = yt * diagonal + ya
zg = zt * ha + za
lg = np.exp(lt) * la
wg = np.exp(wt) * wa
hg = np.exp(ht) * ha
rg = rt + ra
zg = zg - hg / 2
return np.concatenate([xg, yg, zg, wg, lg, hg, rg], axis=-1)
def __init__(self, code_size=7):
super(DeltaXYZWLHRBBoxCoder, self).__init__()
self.code_size = code_size
@staticmethod
def encode_torch(anchors, boxes, means, stds):
def encode(anchors, boxes):
"""
:param boxes: (N, 7+n) x, y, z, w, l, h, r, velo*
:param anchors: (N, 7+n)
......@@ -85,7 +40,7 @@ class Residual3DBoxCoder(object):
return torch.cat([xt, yt, zt, wt, lt, ht, rt, *cts], dim=-1)
@staticmethod
def decode_torch(anchors, box_encodings, means, stds):
def decode(anchors, box_encodings):
"""
:param box_encodings: (N, 7 + n) x, y, z, w, l, h, r
:param anchors: (N, 7)
......
mmdet3d/core/bbox/iou_calculators/iou3d_calculator.py
View file @ 4040dbda
from mmdet3d.ops.iou3d import boxes_iou3d_gpu
from mmdet.core.bbox import bbox_overlaps
from mmdet.core.bbox.iou_calculators.registry import IOU_CALCULATORS
from mmdet.core.bbox.iou_calculators.builder import IOU_CALCULATORS
from .. import box_torch_ops
......
mmdet3d/datasets/__init__.py
View file @ 4040dbda
from mmdet.datasets.registry import DATASETS
from mmdet.datasets.builder import DATASETS
from .builder import build_dataset
from .dataset_wrappers import RepeatFactorDataset
from .kitti2d_dataset import Kitti2DDataset
from .kitti_dataset import KittiDataset
from .loader import DistributedGroupSampler, GroupSampler, build_dataloader
from .nuscenes2d_dataset import NuScenes2DDataset
from .nuscenes_dataset import NuScenesDataset
from .pipelines import (GlobalRotScale, ObjectNoise, ObjectRangeFilter,
ObjectSample, PointShuffle, PointsRangeFilter,
RandomFlip3D)
__all__ = [
'KittiDataset', 'GroupSampler', 'DistributedGroupSampler',
'build_dataloader', 'RepeatFactorDataset', 'DATASETS', 'build_dataset',
'CocoDataset', 'Kitti2DDataset', 'NuScenesDataset', 'NuScenes2DDataset'
'CocoDataset', 'Kitti2DDataset', 'NuScenesDataset', 'ObjectSample',
'RandomFlip3D', 'ObjectNoise', 'GlobalRotScale', 'PointShuffle',
'ObjectRangeFilter', 'PointsRangeFilter', 'Collect3D'
]
mmdet3d/datasets/builder.py
View file @ 4040dbda
import copy
from mmcv.utils import build_from_cfg
from mmdet.datasets import DATASETS, ConcatDataset, RepeatDataset
from mmdet.utils import build_from_cfg
from .dataset_wrappers import RepeatFactorDataset
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment