MMDetection3D源码学习

配置文件

fcos3d_r101_caffe_fpn_gn-head_dcn_2x8_1x_nus-mono3d.py

模型

model = dict(
    backbone=dict(
        dcn=dict(type='DCNv2', deform_groups=1, fallback_on_stride=False),
        stage_with_dcn=(False, False, True, True)))

    class_names = [
        'car', 'truck', 'trailer', 'bus', 'construction_vehicle', 'bicycle',
        'motorcycle', 'pedestrian', 'traffic_cone', 'barrier']
)

数据集

mg_norm_cfg = dict(
    mean=[103.530, 116.280, 123.675],
    std=[1.0, 1.0, 1.0],
    to_rgb=False
)

训练pipeline

train_pipeline = [
    # 从Mono3D文件中加载图像
    dict(type='LoadImageFromFileMono3D'),
    # 加载3D标注
    dict(
        type='LoadAnnotations3D',
        with_bbox=True,
        with_label=True,
        with_attr_label=True,
        with_bbox_3d=True,
        with_label_3d=True,
        with_bbox_depth=True),
    # 缩放训练
    dict(type='Resize', img_scale=(1600, 900), keep_ratio=True),
    dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),
    dict(type='Normalize', **img_norm_cfg),
    dict(type='Pad', size_divisor=32),
    dict(type='DefaultFormatBundle3D', class_names=class_names),
    dict(
        type='Collect3D',
        keys=[
            'img', 'gt_bboxes', 'gt_labels', 'attr_labels', 'gt_bboxes_3d',
            'gt_labels_3d', 'centers2d', 'depths'
        ]),
]

测试pipeline

test_pipeline = [
    dict(type='LoadImageFromFileMono3D'),
    dict(
        type='MultiScaleFlipAug',
        scale_factor=1.0,
        flip=False,
        transforms=[
            dict(type='RandomFlip3D'),
            dict(type='Normalize', **img_norm_cfg),
            dict(type='Pad', size_divisor=32),
            dict(
                 type='DefaultFormatBundle3D',
                 class_names=class_names,
                 with_label=False),
            dict(type='Collect3D', keys=['img']),]
        )
]

数据

data = dict(
    samples_per_gpu=2,
    workers_per_gpu=2,
    train=dict(pipeline=train_pipeline),
    val=dict(pipeline=test_pipeline),
    test=dict(pipeline=test_pipeline)
)
# optimizer
optimizer = dict(
    lr=0.002, paramwise_cfg=dict(bias_lr_mult=2., bias_decay_mult=0.)
)
optimizer_config = dict(
    _delete_=True, grad_clip=dict(max_norm=35, norm_type=2)
)
# learning policy
lr_config = dict(
    policy='step',
    warmup='linear',
    warmup_iters=500,
    warmup_ratio=1.0 / 3,
    step=[8, 11]
)
total_epochs = 12
evaluation = dict(interval=2)