Skip to content

GitLab

Unverified Commit 6c03a971 authored by Tai-Wang's avatar Tai-Wang Committed by GitHub
Browse files

Release v1.1.0rc1 

Release v1.1.0rc1
parents 9611c2d0 ca42c312
Show whitespace changes
Inline Side-by-side
Showing with 798 additions and 341 deletions
configs/_base_/default_runtime.py
View file @ 6c03a971
......@@ -4,7 +4,7 @@ default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=1),
checkpoint=dict(type='CheckpointHook', interval=-1),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='Det3DVisualizationHook'))
......@@ -14,9 +14,6 @@ env_cfg = dict(
dist_cfg=dict(backend='nccl'),
)
vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
type='Det3DLocalVisualizer', vis_backends=vis_backends, name='visualizer')
log_processor = dict(type='LogProcessor', window_size=50, by_epoch=True)
log_level = 'INFO'
......
configs/_base_/models/multiview_dfm.py 0 → 100644
View file @ 6c03a971
model = dict(
type='MultiViewDfM',
data_preprocessor=dict(
type='Det3DDataPreprocessor',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
bgr_to_rgb=True,
pad_size_divisor=32),
backbone=dict(
type='mmdet.ResNet',
depth=101,
num_stages=4,
out_indices=(0, 1, 2, 3),
frozen_stages=1,
norm_cfg=dict(type='BN', requires_grad=False),
norm_eval=True,
style='pytorch',
init_cfg=dict(type='Pretrained', checkpoint='torchvision://resnet101'),
dcn=dict(type='DCNv2', deform_groups=1, fallback_on_stride=False),
stage_with_dcn=(False, False, True, True)),
neck=dict(
type='mmdet.FPN',
in_channels=[256, 512, 1024, 2048],
out_channels=64,
num_outs=4),
neck_2d=None,
bbox_head_2d=None,
backbone_stereo=None,
depth_head=None,
backbone_3d=None,
neck_3d=dict(type='OutdoorImVoxelNeck', in_channels=64, out_channels=256),
valid_sample=True,
voxel_size=(0.5, 0.5, 0.5), # n_voxels=[240, 300, 12]
anchor_generator=dict(
type='AlignedAnchor3DRangeGenerator',
ranges=[[-35.0, -75.0, -2, 75.0, 75.0, 4]],
rotations=[.0]),
bbox_head=dict(
type='Anchor3DHead',
num_classes=3,
in_channels=256,
feat_channels=256,
use_direction_classifier=True,
anchor_generator=dict(
type='AlignedAnchor3DRangeGenerator',
ranges=[[-35.0, -75.0, -0.0345, 75.0, 75.0, -0.0345],
[-35.0, -75.0, 0, 75.0, 75.0, 0],
[-35.0, -75.0, -0.1188, 75.0, 75.0, -0.1188]],
sizes=[
[4.73, 2.08, 1.77], # car
[0.91, 0.84, 1.74], # pedestrian
[1.81, 0.84, 1.77], # cyclist
],
rotations=[0, 1.57],
reshape_out=False),
diff_rad_by_sin=True,
dir_offset=-0.7854, # -pi / 4
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='mmdet.FocalLoss',
use_sigmoid=True,
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
loss_bbox=dict(
type='mmdet.SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_dir=dict(
type='mmdet.CrossEntropyLoss', use_sigmoid=False,
loss_weight=0.2)),
train_cfg=dict(
assigner=[
dict( # for Car
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.6,
neg_iou_thr=0.45,
min_pos_iou=0.45,
ignore_iof_thr=-1),
dict( # for Pedestrian
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Cyclist
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
],
allowed_border=0,
pos_weight=-1,
debug=False),
test_cfg=dict(
use_rotate_nms=True,
nms_across_levels=False,
nms_thr=0.05,
score_thr=0.001,
min_bbox_size=0,
nms_pre=500,
max_num=100))
configs/_base_/models/point_rcnn.py
View file @ 6c03a971
model = dict(
type='PointRCNN',
data_preprocessor=dict(type='Det3DDataPreprocessor'),
backbone=dict(
type='PointNet2SAMSG',
in_channels=4,
......@@ -34,14 +35,14 @@ model = dict(
cls_linear_channels=(256, 256),
reg_linear_channels=(256, 256)),
cls_loss=dict(
type='FocalLoss',
type='mmdet.FocalLoss',
use_sigmoid=True,
reduction='sum',
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
bbox_loss=dict(
type='SmoothL1Loss',
type='mmdet.SmoothL1Loss',
beta=1.0 / 9.0,
reduction='sum',
loss_weight=1.0),
......@@ -55,12 +56,22 @@ model = dict(
1.73]])),
roi_head=dict(
type='PointRCNNRoIHead',
point_roi_extractor=dict(
bbox_roi_extractor=dict(
type='Single3DRoIPointExtractor',
roi_layer=dict(type='RoIPointPool3d', num_sampled_points=512)),
bbox_head=dict(
type='PointRCNNBboxHead',
num_classes=1,
loss_bbox=dict(
type='mmdet.SmoothL1Loss',
beta=1.0 / 9.0,
reduction='sum',
loss_weight=1.0),
loss_cls=dict(
type='mmdet.CrossEntropyLoss',
use_sigmoid=True,
reduction='sum',
loss_weight=1.0),
pred_layer_cfg=dict(
in_channels=512,
cls_conv_channels=(256, 256),
......@@ -79,13 +90,16 @@ model = dict(
train_cfg=dict(
pos_distance_thr=10.0,
rpn=dict(
nms_cfg=dict(
use_rotate_nms=True, iou_thr=0.8, nms_pre=9000, nms_post=512),
score_thr=None),
rpn_proposal=dict(
use_rotate_nms=True,
score_thr=None,
iou_thr=0.8,
nms_pre=9000,
nms_post=512)),
rcnn=dict(
assigner=[
dict( # for Car
type='MaxIoUAssigner',
dict( # for Pedestrian
type='Max3DIoUAssigner',
iou_calculator=dict(
type='BboxOverlaps3D', coordinate='lidar'),
pos_iou_thr=0.55,
......@@ -93,8 +107,8 @@ model = dict(
min_pos_iou=0.55,
ignore_iof_thr=-1,
match_low_quality=False),
dict( # for Pedestrian
type='MaxIoUAssigner',
dict( # for Cyclist
type='Max3DIoUAssigner',
iou_calculator=dict(
type='BboxOverlaps3D', coordinate='lidar'),
pos_iou_thr=0.55,
......@@ -102,8 +116,8 @@ model = dict(
min_pos_iou=0.55,
ignore_iof_thr=-1,
match_low_quality=False),
dict( # for Cyclist
type='MaxIoUAssigner',
dict( # for Car
type='Max3DIoUAssigner',
iou_calculator=dict(
type='BboxOverlaps3D', coordinate='lidar'),
pos_iou_thr=0.55,
......@@ -126,6 +140,9 @@ model = dict(
test_cfg=dict(
rpn=dict(
nms_cfg=dict(
use_rotate_nms=True, iou_thr=0.85, nms_pre=9000, nms_post=512),
score_thr=None),
use_rotate_nms=True,
iou_thr=0.85,
nms_pre=9000,
nms_post=512,
score_thr=None)),
rcnn=dict(use_rotate_nms=True, nms_thr=0.1, score_thr=0.1)))
configs/_base_/schedules/cosine.py
View file @ 6c03a971
......@@ -22,3 +22,9 @@ param_scheduler = [
train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=40, val_interval=1)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (8 GPUs) x (2 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=16)
configs/_base_/schedules/cyclic-20e.py
View file @ 6c03a971
......@@ -57,3 +57,9 @@ param_scheduler = [
train_cfg = dict(by_epoch=True, max_epochs=20, val_interval=20)
val_cfg = dict()
test_cfg = dict()
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (8 GPUs) x (4 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=32)
configs/_base_/schedules/cyclic-40e.py
View file @ 6c03a971
......@@ -59,3 +59,9 @@ param_scheduler = [
train_cfg = dict(by_epoch=True, max_epochs=40, val_interval=1)
val_cfg = dict()
test_cfg = dict()
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (8 GPUs) x (6 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=48)
configs/_base_/schedules/mmdet-schedule-1x.py
View file @ 6c03a971
......@@ -20,3 +20,9 @@ param_scheduler = [
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='SGD', lr=0.02, momentum=0.9, weight_decay=0.0001))
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (8 GPUs) x (2 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=16)
configs/_base_/schedules/schedule-2x.py
View file @ 6c03a971
......@@ -28,3 +28,9 @@ param_scheduler = [
milestones=[20, 23],
gamma=0.1)
]
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (8 GPUs) x (4 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=32)
configs/_base_/schedules/schedule-3x.py
View file @ 6c03a971
......@@ -23,3 +23,9 @@ param_scheduler = [
milestones=[24, 32],
gamma=0.1)
]
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (4 GPUs) x (8 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=32)
configs/_base_/schedules/seg-cosine-100e.py
View file @ 6c03a971
......@@ -19,3 +19,9 @@ param_scheduler = [
train_cfg = dict(by_epoch=True, max_epochs=100)
val_cfg = dict(interval=1)
test_cfg = dict()
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (4 GPUs) x (32 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=128)
configs/_base_/schedules/seg-cosine-150e.py
View file @ 6c03a971
......@@ -19,3 +19,9 @@ param_scheduler = [
train_cfg = dict(by_epoch=True, max_epochs=150)
val_cfg = dict(interval=1)
test_cfg = dict()
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (8 GPUs) x (8 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=64)
configs/_base_/schedules/seg-cosine-200e.py
View file @ 6c03a971
......@@ -19,3 +19,9 @@ param_scheduler = [
train_cfg = dict(by_epoch=True, max_epochs=200)
val_cfg = dict(interval=1)
test_cfg = dict()
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (2 GPUs) x (16 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=32)
configs/_base_/schedules/seg-cosine-50e.py
View file @ 6c03a971
......@@ -19,3 +19,9 @@ param_scheduler = [
train_cfg = dict(by_epoch=True, max_epochs=50)
val_cfg = dict(interval=1)
test_cfg = dict()
# Default setting for scaling LR automatically
# - `enable` means enable scaling LR automatically
# or not by default.
# - `base_batch_size` = (2 GPUs) x (16 samples per GPU).
auto_scale_lr = dict(enable=False, base_batch_size=32)
configs/benchmark/hv_PartA2_secfpn_4x8_cyclic_80e_pcdet_kitti-3d-3class.py
View file @ 6c03a971
......@@ -48,14 +48,16 @@ model = dict(
assign_per_class=True,
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
type='mmdet.FocalLoss',
use_sigmoid=True,
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
loss_bbox=dict(type='SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_bbox=dict(
type='mmdet.SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_dir=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.2)),
type='mmdet.CrossEntropyLoss', use_sigmoid=False,
loss_weight=0.2)),
roi_head=dict(
type='PartAggregationROIHead',
num_classes=3,
......@@ -66,14 +68,16 @@ model = dict(
seg_score_thr=0.3,
num_classes=3,
loss_seg=dict(
type='FocalLoss',
type='mmdet.FocalLoss',
use_sigmoid=True,
reduction='sum',
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
loss_part=dict(
type='CrossEntropyLoss', use_sigmoid=True, loss_weight=1.0)),
type='mmdet.CrossEntropyLoss',
use_sigmoid=True,
loss_weight=1.0)),
seg_roi_extractor=dict(
type='Single3DRoIAwareExtractor',
roi_layer=dict(
......@@ -81,7 +85,7 @@ model = dict(
out_size=14,
max_pts_per_voxel=128,
mode='max')),
part_roi_extractor=dict(
bbox_roi_extractor=dict(
type='Single3DRoIAwareExtractor',
roi_layer=dict(
type='RoIAwarePool3d',
......@@ -105,12 +109,12 @@ model = dict(
roi_feat_size=14,
with_corner_loss=True,
loss_bbox=dict(
type='SmoothL1Loss',
type='mmdet.SmoothL1Loss',
beta=1.0 / 9.0,
reduction='sum',
loss_weight=1.0),
loss_cls=dict(
type='CrossEntropyLoss',
type='mmdet.CrossEntropyLoss',
use_sigmoid=True,
reduction='sum',
loss_weight=1.0))),
......@@ -119,21 +123,21 @@ model = dict(
rpn=dict(
assigner=[
dict( # for Pedestrian
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Cyclist
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Car
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.6,
neg_iou_thr=0.45,
......@@ -153,7 +157,7 @@ model = dict(
rcnn=dict(
assigner=[
dict( # for Pedestrian
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(
type='BboxOverlaps3D', coordinate='lidar'),
pos_iou_thr=0.55,
......@@ -161,7 +165,7 @@ model = dict(
min_pos_iou=0.55,
ignore_iof_thr=-1),
dict( # for Cyclist
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(
type='BboxOverlaps3D', coordinate='lidar'),
pos_iou_thr=0.55,
......@@ -169,7 +173,7 @@ model = dict(
min_pos_iou=0.55,
ignore_iof_thr=-1),
dict( # for Car
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(
type='BboxOverlaps3D', coordinate='lidar'),
pos_iou_thr=0.55,
......@@ -200,12 +204,13 @@ model = dict(
use_rotate_nms=True,
use_raw_score=True,
nms_thr=0.01,
score_thr=0.3)))
score_thr=0.1)))
# dataset settings
dataset_type = 'KittiDataset'
data_root = 'data/kitti/'
class_names = ['Pedestrian', 'Cyclist', 'Car']
metainfo = dict(CLASSES=class_names)
input_modality = dict(use_lidar=True, use_camera=False)
db_sampler = dict(
data_root=data_root,
......@@ -215,9 +220,8 @@ db_sampler = dict(
filter_by_difficulty=[-1],
filter_by_min_points=dict(Car=5, Pedestrian=5, Cyclist=5)),
classes=class_names,
sample_groups=dict(Car=20, Pedestrian=15, Cyclist=15),
points_loader=dict(
type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4))
sample_groups=dict(Car=20, Pedestrian=15, Cyclist=15))
train_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True),
......@@ -231,8 +235,9 @@ train_pipeline = [
dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
dict(type='ObjectNameFilter', classes=class_names),
dict(type='PointShuffle'),
dict(type='DefaultFormatBundle3D', class_names=class_names),
dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_labels_3d', 'gt_bboxes_3d'])
]
test_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
......@@ -249,88 +254,133 @@ test_pipeline = [
translation_std=[0, 0, 0]),
dict(type='RandomFlip3D'),
dict(
type='PointsRangeFilter', point_cloud_range=point_cloud_range),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
])
type='PointsRangeFilter', point_cloud_range=point_cloud_range)
]),
dict(type='Pack3DDetInputs', keys=['points'])
]
# construct a pipeline for data and gt loading in show function
# please keep its loading function consistent with test_pipeline (e.g. client)
eval_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
dict(type='Pack3DDetInputs', keys=['points'])
]
data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
train_dataloader = dict(
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_train.pkl',
split='training',
pts_prefix='velodyne_reduced',
ann_file='kitti_infos_train.pkl',
data_prefix=dict(pts='training/velodyne_reduced'),
pipeline=train_pipeline,
modality=input_modality,
classes=class_names,
test_mode=False),
val=dict(
test_mode=False,
metainfo=metainfo,
# we use box_type_3d='LiDAR' in kitti and nuscenes dataset
# and box_type_3d='Depth' in sunrgbd and scannet dataset.
box_type_3d='LiDAR'))
val_dataloader = dict(
batch_size=1,
num_workers=1,
persistent_workers=True,
drop_last=False,
sampler=dict(type='DefaultSampler', shuffle=False),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
data_prefix=dict(pts='training/velodyne_reduced'),
ann_file='kitti_infos_val.pkl',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True),
test=dict(
type=dataset_type,
data_root=data_root,
test_mode=True,
metainfo=metainfo,
box_type_3d='LiDAR'))
test_dataloader = val_dataloader
val_evaluator = dict(
type='KittiMetric',
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True))
metric='bbox')
test_evaluator = val_evaluator
# optimizer
lr = 0.001 # max learning rate
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),
cyclic_times=1,
step_ratio_up=0.4)
momentum_config = dict(
policy='cyclic',
target_ratio=(0.85 / 0.95, 1),
cyclic_times=1,
step_ratio_up=0.4)
checkpoint_config = dict(interval=1)
evaluation = dict(interval=1, pipeline=eval_pipeline)
# yapf:disable
log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
dict(type='TensorboardLoggerHook')
])
# yapf:enable
# runtime settings
runner = dict(type='EpochBasedRunner', max_epochs=80)
dist_params = dict(backend='nccl', port=29506)
epoch_num = 80
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=lr, betas=(0.95, 0.99), weight_decay=0.01),
clip_grad=dict(max_norm=10, norm_type=2))
# learning policy
param_scheduler = [
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.4,
eta_min=lr * 10,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.6,
eta_min=lr * 1e-4,
begin=epoch_num * 0.4,
end=epoch_num * 1,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.4,
eta_min=0.85 / 0.95,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.6,
eta_min=1,
begin=epoch_num * 0.4,
end=epoch_num * 1,
convert_to_iter_based=True)
]
train_cfg = dict(by_epoch=True, max_epochs=epoch_num, val_interval=50)
val_cfg = dict()
test_cfg = dict()
auto_scale_lr = dict(enable=False, base_batch_size=32)
default_scope = 'mmdet3d'
default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=1),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='Det3DVisualizationHook'))
custom_hooks = [
dict(type='BenchmarkHook'),
]
env_cfg = dict(
cudnn_benchmark=False,
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
dist_cfg=dict(backend='nccl'),
)
vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
type='Det3DLocalVisualizer', vis_backends=vis_backends, name='visualizer')
log_processor = dict(type='LogProcessor', window_size=50, by_epoch=True)
log_level = 'INFO'
load_from = None
resume = False
find_unused_parameters = True
work_dir = './work_dirs/parta2_secfpn_80e'
load_from = None
resume_from = None
workflow = [('train', 1)]
configs/benchmark/hv_pointpillars_secfpn_3x8_100e_det3d_kitti-3d-car.py
View file @ 6c03a971
......@@ -46,18 +46,20 @@ model = dict(
diff_rad_by_sin=True,
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
type='mmdet.FocalLoss',
use_sigmoid=True,
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
loss_bbox=dict(type='SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_bbox=dict(
type='mmdet.SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_dir=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.2)),
type='mmdet.CrossEntropyLoss', use_sigmoid=False,
loss_weight=0.2)),
# model training and testing settings
train_cfg=dict(
assigner=dict(
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.6,
neg_iou_thr=0.45,
......@@ -79,6 +81,7 @@ model = dict(
dataset_type = 'KittiDataset'
data_root = 'data/kitti/'
class_names = ['Car']
metainfo = dict(CLASSES=class_names)
input_modality = dict(use_lidar=True, use_camera=False)
db_sampler = dict(
data_root=data_root,
......@@ -86,9 +89,7 @@ db_sampler = dict(
rate=1.0,
prepare=dict(filter_by_difficulty=[-1], filter_by_min_points=dict(Car=5)),
sample_groups=dict(Car=15),
classes=class_names,
points_loader=dict(
type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4))
classes=class_names)
train_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
......@@ -108,99 +109,140 @@ train_pipeline = [
dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
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_3d', 'gt_labels_3d'])
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_labels_3d', 'gt_bboxes_3d'])
]
test_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
dict(type='Pack3DDetInputs', keys=['points'])
]
# construct a pipeline for data and gt loading in show function
# please keep its loading function consistent with test_pipeline (e.g. client)
eval_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
dict(type='Pack3DDetInputs', keys=['points'])
]
data = dict(
samples_per_gpu=3,
workers_per_gpu=3,
train=dict(
train_dataloader = dict(
batch_size=3,
num_workers=3,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=True),
dataset=dict(
type='RepeatDataset',
times=2,
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_train.pkl',
split='training',
pts_prefix='velodyne_reduced',
ann_file='kitti_infos_train.pkl',
data_prefix=dict(pts='training/velodyne_reduced'),
pipeline=train_pipeline,
modality=input_modality,
classes=class_names,
test_mode=False)),
val=dict(
test_mode=False,
metainfo=metainfo,
# we use box_type_3d='LiDAR' in kitti and nuscenes dataset
# and box_type_3d='Depth' in sunrgbd and scannet dataset.
box_type_3d='LiDAR')))
val_dataloader = dict(
batch_size=1,
num_workers=1,
persistent_workers=True,
drop_last=False,
sampler=dict(type='DefaultSampler', shuffle=False),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
data_prefix=dict(pts='training/velodyne_reduced'),
ann_file='kitti_infos_val.pkl',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True),
test=dict(
type=dataset_type,
data_root=data_root,
test_mode=True,
metainfo=metainfo,
box_type_3d='LiDAR'))
test_dataloader = val_dataloader
val_evaluator = dict(
type='KittiMetric',
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True))
metric='bbox')
test_evaluator = val_evaluator
# optimizer
lr = 0.001 # max learning rate
optimizer = dict(
type='AdamW',
lr=lr,
betas=(0.95, 0.99), # the momentum is change during training
weight_decay=0.01)
optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
epoch_num = 50
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=lr, betas=(0.95, 0.99), weight_decay=0.01),
clip_grad=dict(max_norm=10, norm_type=2))
# learning policy
lr_config = dict(
policy='cyclic',
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),
cyclic_times=1,
step_ratio_up=0.4)
checkpoint_config = dict(interval=1)
evaluation = dict(interval=1, pipeline=eval_pipeline)
# yapf:disable
log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
dict(type='TensorboardLoggerHook')
])
# yapf:enable
# runtime settings
runner = dict(type='EpochBasedRunner', max_epochs=50)
dist_params = dict(backend='nccl')
param_scheduler = [
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.4,
eta_min=lr * 10,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.6,
eta_min=lr * 1e-4,
begin=epoch_num * 0.4,
end=epoch_num * 1,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.4,
eta_min=0.85 / 0.95,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.6,
eta_min=1,
begin=epoch_num * 0.4,
end=epoch_num * 1,
convert_to_iter_based=True)
]
train_cfg = dict(by_epoch=True, max_epochs=epoch_num, val_interval=50)
val_cfg = dict()
test_cfg = dict()
auto_scale_lr = dict(enable=False, base_batch_size=24)
default_scope = 'mmdet3d'
default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=1),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='Det3DVisualizationHook'))
custom_hooks = [
dict(type='BenchmarkHook'),
]
env_cfg = dict(
cudnn_benchmark=False,
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
dist_cfg=dict(backend='nccl'),
)
vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
type='Det3DLocalVisualizer', vis_backends=vis_backends, name='visualizer')
log_processor = dict(type='LogProcessor', window_size=50, by_epoch=True)
log_level = 'INFO'
work_dir = './work_dirs/pp_secfpn_100e'
load_from = None
resume_from = None
workflow = [('train', 50)]
resume = False
work_dir = './work_dirs/pp_secfpn_100e'
configs/benchmark/hv_pointpillars_secfpn_4x8_80e_pcdet_kitti-3d-3class.py
View file @ 6c03a971
......@@ -56,34 +56,35 @@ model = dict(
diff_rad_by_sin=True,
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
type='mmdet.FocalLoss',
use_sigmoid=True,
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
loss_bbox=dict(type='SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_bbox=dict(
type='mmdet.SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_dir=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.2),
type='mmdet.CrossEntropyLoss', use_sigmoid=False, loss_weight=0.2),
),
# model training and testing settings
train_cfg=dict(
assigner=[
dict( # for Pedestrian
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Cyclist
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Car
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.6,
neg_iou_thr=0.45,
......@@ -106,6 +107,8 @@ model = dict(
dataset_type = 'KittiDataset'
data_root = 'data/kitti/'
class_names = ['Pedestrian', 'Cyclist', 'Car']
metainfo = dict(CLASSES=class_names)
input_modality = dict(use_lidar=True, use_camera=False)
db_sampler = dict(
data_root=data_root,
......@@ -123,9 +126,7 @@ db_sampler = dict(
Car=15,
Pedestrian=15,
Cyclist=15,
),
points_loader=dict(
type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4))
))
train_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
......@@ -139,8 +140,9 @@ train_pipeline = [
dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
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_3d', 'gt_labels_3d']),
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_labels_3d', 'gt_bboxes_3d'])
]
test_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
......@@ -158,91 +160,132 @@ test_pipeline = [
dict(type='RandomFlip3D'),
dict(
type='PointsRangeFilter', point_cloud_range=point_cloud_range),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
])
]),
dict(type='Pack3DDetInputs', keys=['points'])
]
# construct a pipeline for data and gt loading in show function
# please keep its loading function consistent with test_pipeline (e.g. client)
eval_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
dict(type='Pack3DDetInputs', keys=['points'])
]
data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
train_dataloader = dict(
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_train.pkl',
split='training',
pts_prefix='velodyne_reduced',
ann_file='kitti_infos_train.pkl',
data_prefix=dict(pts='training/velodyne_reduced'),
pipeline=train_pipeline,
modality=input_modality,
classes=class_names,
test_mode=False),
val=dict(
test_mode=False,
metainfo=metainfo,
# we use box_type_3d='LiDAR' in kitti and nuscenes dataset
# and box_type_3d='Depth' in sunrgbd and scannet dataset.
box_type_3d='LiDAR'))
val_dataloader = dict(
batch_size=1,
num_workers=1,
persistent_workers=True,
drop_last=False,
sampler=dict(type='DefaultSampler', shuffle=False),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
data_prefix=dict(pts='training/velodyne_reduced'),
ann_file='kitti_infos_val.pkl',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True),
test=dict(
type=dataset_type,
data_root=data_root,
test_mode=True,
metainfo=metainfo,
box_type_3d='LiDAR'))
test_dataloader = val_dataloader
val_evaluator = dict(
type='KittiMetric',
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True))
metric='bbox')
test_evaluator = val_evaluator
# optimizer
lr = 0.0003 # max learning rate
optimizer = dict(
type='AdamW',
lr=lr,
betas=(0.95, 0.99), # the momentum is change during training
weight_decay=0.01)
optimizer_config = dict(grad_clip=dict(max_norm=10, norm_type=2))
epoch_num = 80
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=lr, betas=(0.95, 0.99), weight_decay=0.01),
clip_grad=dict(max_norm=10, norm_type=2))
# learning policy
lr_config = dict(
policy='cyclic',
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),
cyclic_times=1,
step_ratio_up=0.4)
checkpoint_config = dict(interval=1)
evaluation = dict(interval=2, pipeline=eval_pipeline)
# yapf:disable
log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
dict(type='TensorboardLoggerHook')
])
# yapf:enable
# runtime settings
runner = dict(type='EpochBasedRunner', max_epochs=80)
dist_params = dict(backend='nccl')
param_scheduler = [
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.4,
eta_min=lr * 10,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.6,
eta_min=lr * 1e-4,
begin=epoch_num * 0.4,
end=epoch_num * 1,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.4,
eta_min=0.85 / 0.95,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.6,
eta_min=1,
begin=epoch_num * 0.4,
end=epoch_num * 1,
convert_to_iter_based=True)
]
train_cfg = dict(by_epoch=True, max_epochs=epoch_num, val_interval=50)
val_cfg = dict()
test_cfg = dict()
auto_scale_lr = dict(enable=False, base_batch_size=32)
default_scope = 'mmdet3d'
default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=1),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='Det3DVisualizationHook'))
custom_hooks = [
dict(type='BenchmarkHook'),
]
env_cfg = dict(
cudnn_benchmark=False,
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
dist_cfg=dict(backend='nccl'),
)
vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
type='Det3DLocalVisualizer', vis_backends=vis_backends, name='visualizer')
log_processor = dict(type='LogProcessor', window_size=50, by_epoch=True)
log_level = 'INFO'
work_dir = './work_dirs/pp_secfpn_80e'
load_from = None
resume_from = None
workflow = [('train', 1)]
resume = False
work_dir = './work_dirs/pp_secfpn_80e'
configs/benchmark/hv_second_secfpn_4x8_80e_pcdet_kitti-3d-3class.py
View file @ 6c03a971
......@@ -48,33 +48,35 @@ model = dict(
diff_rad_by_sin=True,
bbox_coder=dict(type='DeltaXYZWLHRBBoxCoder'),
loss_cls=dict(
type='FocalLoss',
type='mmdet.FocalLoss',
use_sigmoid=True,
gamma=2.0,
alpha=0.25,
loss_weight=1.0),
loss_bbox=dict(type='SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_bbox=dict(
type='mmdet.SmoothL1Loss', beta=1.0 / 9.0, loss_weight=2.0),
loss_dir=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.2)),
type='mmdet.CrossEntropyLoss', use_sigmoid=False,
loss_weight=0.2)),
# model training and testing settings
train_cfg=dict(
assigner=[
dict( # for Pedestrian
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Cyclist
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.5,
neg_iou_thr=0.35,
min_pos_iou=0.35,
ignore_iof_thr=-1),
dict( # for Car
type='MaxIoUAssigner',
type='Max3DIoUAssigner',
iou_calculator=dict(type='BboxOverlapsNearest3D'),
pos_iou_thr=0.6,
neg_iou_thr=0.45,
......@@ -97,7 +99,8 @@ model = dict(
dataset_type = 'KittiDataset'
data_root = 'data/kitti/'
class_names = ['Pedestrian', 'Cyclist', 'Car']
input_modality = dict(use_lidar=False, use_camera=False)
metainfo = dict(CLASSES=class_names)
input_modality = dict(use_lidar=True, use_camera=False)
db_sampler = dict(
data_root=data_root,
info_path=data_root + 'kitti_dbinfos_train.pkl',
......@@ -114,12 +117,7 @@ db_sampler = dict(
Car=20,
Pedestrian=15,
Cyclist=15,
),
points_loader=dict(
type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4))
file_client_args = dict(backend='disk')
# file_client_args = dict(
# backend='petrel', path_mapping=dict(data='s3://kitti_data/'))
))
train_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
......@@ -133,8 +131,9 @@ train_pipeline = [
dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
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_3d', 'gt_labels_3d'])
dict(
type='Pack3DDetInputs',
keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
]
test_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
......@@ -151,87 +150,132 @@ test_pipeline = [
translation_std=[0, 0, 0]),
dict(type='RandomFlip3D'),
dict(
type='PointsRangeFilter', point_cloud_range=point_cloud_range),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
])
type='PointsRangeFilter', point_cloud_range=point_cloud_range)
]),
dict(type='Pack3DDetInputs', keys=['points'])
]
# construct a pipeline for data and gt loading in show function
# please keep its loading function consistent with test_pipeline (e.g. client)
eval_pipeline = [
dict(type='LoadPointsFromFile', coord_type='LIDAR', load_dim=4, use_dim=4),
dict(
type='DefaultFormatBundle3D',
class_names=class_names,
with_label=False),
dict(type='Collect3D', keys=['points'])
dict(type='Pack3DDetInputs', keys=['points'])
]
data = dict(
samples_per_gpu=4,
workers_per_gpu=4,
train=dict(
train_dataloader = dict(
batch_size=4,
num_workers=4,
persistent_workers=True,
sampler=dict(type='DefaultSampler', shuffle=True),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_train.pkl',
split='training',
pts_prefix='velodyne_reduced',
ann_file='kitti_infos_train.pkl',
data_prefix=dict(pts='training/velodyne_reduced'),
pipeline=train_pipeline,
modality=input_modality,
classes=class_names,
test_mode=False),
val=dict(
test_mode=False,
metainfo=metainfo,
# we use box_type_3d='LiDAR' in kitti and nuscenes dataset
# and box_type_3d='Depth' in sunrgbd and scannet dataset.
box_type_3d='LiDAR'))
val_dataloader = dict(
batch_size=1,
num_workers=1,
persistent_workers=True,
drop_last=False,
sampler=dict(type='DefaultSampler', shuffle=False),
dataset=dict(
type=dataset_type,
data_root=data_root,
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
data_prefix=dict(pts='training/velodyne_reduced'),
ann_file='kitti_infos_val.pkl',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True),
test=dict(
type=dataset_type,
data_root=data_root,
test_mode=True,
metainfo=metainfo,
box_type_3d='LiDAR'))
test_dataloader = val_dataloader
val_evaluator = dict(
type='KittiMetric',
ann_file=data_root + 'kitti_infos_val.pkl',
split='training',
pts_prefix='velodyne_reduced',
pipeline=test_pipeline,
modality=input_modality,
classes=class_names,
test_mode=True))
metric='bbox')
test_evaluator = val_evaluator
# optimizer
lr = 0.0003 # max learning rate
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),
cyclic_times=1,
step_ratio_up=0.4)
momentum_config = dict(
policy='cyclic',
target_ratio=(0.85 / 0.95, 1),
cyclic_times=1,
step_ratio_up=0.4)
checkpoint_config = dict(interval=1)
evaluation = dict(interval=2, pipeline=eval_pipeline)
# yapf:disable
log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
dict(type='TensorboardLoggerHook')
])
# yapf:enable
# runtime settings
runner = dict(type='EpochBasedRunner', max_epochs=80)
dist_params = dict(backend='nccl')
epoch_num = 80
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=lr, betas=(0.95, 0.99), weight_decay=0.01),
clip_grad=dict(max_norm=10, norm_type=2))
# learning policy
param_scheduler = [
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.4,
eta_min=lr * 10,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingLR',
T_max=epoch_num * 0.6,
eta_min=lr * 1e-4,
begin=epoch_num * 0.4,
end=epoch_num * 1,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.4,
eta_min=0.85 / 0.95,
begin=0,
end=epoch_num * 0.4,
by_epoch=True,
convert_to_iter_based=True),
dict(
type='CosineAnnealingMomentum',
T_max=epoch_num * 0.6,
eta_min=1,
begin=epoch_num * 0.4,
end=epoch_num * 1,
convert_to_iter_based=True)
]
train_cfg = dict(by_epoch=True, max_epochs=epoch_num, val_interval=50)
val_cfg = dict()
test_cfg = dict()
auto_scale_lr = dict(enable=False, base_batch_size=32)
default_scope = 'mmdet3d'
default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=1),
sampler_seed=dict(type='DistSamplerSeedHook'),
visualization=dict(type='Det3DVisualizationHook'))
custom_hooks = [
dict(type='BenchmarkHook'),
]
env_cfg = dict(
cudnn_benchmark=False,
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
dist_cfg=dict(backend='nccl'),
)
vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
type='Det3DLocalVisualizer', vis_backends=vis_backends, name='visualizer')
log_processor = dict(type='LogProcessor', window_size=50, by_epoch=True)
log_level = 'INFO'
work_dir = './work_dirs/sec_secfpn_80e'
load_from = None
resume_from = None
workflow = [('train', 1)]
resume = False
work_dir = './work_dirs/pp_secfpn_100e'
configs/dfm/multiview-dfm_r101-dcn_16xb2_waymoD5-3d-3class.py 0 → 100644
View file @ 6c03a971
_base_ = [
'../_base_/datasets/waymoD5-mv3d-3class.py',
'../_base_/models/multiview_dfm.py'
]
# optimizer
optim_wrapper = dict(
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=0.0005, weight_decay=0.0001),
paramwise_cfg=dict(
custom_keys={'backbone': dict(lr_mult=0.1, decay_mult=1.0)}),
clip_grad=dict(max_norm=35., norm_type=2))
param_scheduler = [
dict(
type='MultiStepLR',
begin=0,
end=24,
by_epoch=True,
milestones=[16, 22],
gamma=0.1)
]
# hooks
default_hooks = dict(
timer=dict(type='IterTimerHook'),
logger=dict(type='LoggerHook', interval=50),
param_scheduler=dict(type='ParamSchedulerHook'),
checkpoint=dict(type='CheckpointHook', interval=1, max_keep_ckpts=1),
sampler_seed=dict(type='DistSamplerSeedHook'),
)
# training schedule for 2x
train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=24, val_interval=24)
val_cfg = dict(type='ValLoop')
test_cfg = dict(type='TestLoop')
# runtime
default_scope = 'mmdet3d'
env_cfg = dict(
cudnn_benchmark=False,
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
dist_cfg=dict(backend='nccl'),
)
log_level = 'INFO'
load_from = None
resume = False
find_unused_parameters = True # only 1 of 4 FPN outputs is used
configs/dfm/multiview-dfm_r101-dcn_centerhead_16xb2_waymoD5-3d-3class.py 0 → 100644
View file @ 6c03a971
_base_ = ['./multiview-dfm_r101_dcn_2x16_waymoD5-3d-3class.py']
model = dict(
bbox_head=dict(
_delete_=True,
type='CenterHead',
in_channels=256,
tasks=[
dict(num_class=1, class_names=['Pedestrian']),
dict(num_class=1, class_names=['Cyclist']),
dict(num_class=1, class_names=['Car']),
],
common_heads=dict(reg=(2, 2), height=(1, 2), dim=(3, 2), rot=(2, 2)),
share_conv_channel=64,
bbox_coder=dict(
type='CenterPointBBoxCoder',
post_center_range=[-35.0, -75.0, -2, 75.0, 75.0, 4],
pc_range=[-35.0, -75.0, -2, 75.0, 75.0, 4],
max_num=2000,
score_threshold=0,
out_size_factor=1,
voxel_size=(.50, .50),
code_size=7),
separate_head=dict(
type='SeparateHead', init_bias=-2.19, final_kernel=3),
loss_cls=dict(type='mmdet.GaussianFocalLoss', reduction='mean'),
loss_bbox=dict(
type='mmdet.L1Loss', reduction='mean', loss_weight=0.25),
norm_bbox=True),
train_cfg=dict(
_delete_=True,
grid_size=[220, 300, 1],
voxel_size=(0.5, 0.5, 6),
out_size_factor=1,
dense_reg=1,
gaussian_overlap=0.1,
max_objs=500,
min_radius=2,
code_weights=[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0],
point_cloud_range=[-35.0, -75.0, -2, 75.0, 75.0, 4]),
test_cfg=dict(
_delete_=True,
post_center_limit_range=[-35.0, -75.0, -2, 75.0, 75.0, 4],
max_per_img=4096,
max_pool_nms=False,
min_radius=[0.5, 2, 6],
score_threshold=0,
out_size_factor=1,
voxel_size=(0.5, 0.5),
nms_type='circle',
pre_max_size=2000,
post_max_size=200,
nms_thr=0.2))
configs/fcos3d/fcos3d_r101-caffe-dcn_fpn_head-gn_8xb2-1x_nus-mono3d.py
View file @ 6c03a971
......@@ -53,9 +53,7 @@ test_pipeline = [
]
train_dataloader = dict(
batch_size=2,
num_workers=2,
dataset=dict(dataset=dict(pipeline=train_pipeline)))
batch_size=2, num_workers=2, dataset=dict(pipeline=train_pipeline))
test_dataloader = dict(dataset=dict(pipeline=test_pipeline))
val_dataloader = dict(dataset=dict(pipeline=test_pipeline))
......
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