Pointpillars代码解读——数据处理

最新推荐文章于 2025-06-18 17:46:43 发布

原创

最新推荐文章于 2025-06-18 17:46:43 发布 · 1.8k 阅读

36 ·

CC 4.0 BY-SA版权

文章标签：

#目标检测 #深度学习 #python #cnn #人工智能 #numpy #github

Pointpillars论文：https://arxiv.org/abs/1812.05784

代码：GitHub - open-mmlab/OpenPCDet: OpenPCDet Toolbox for LiDAR-based 3D Object Detection.

算法复现：基于kitti数据集的3D目标检测算法的训练流程_mini kitti 数据集-CSDN博客

文献理解：PointPillars文献理解_pillar feature net-CSDN博客

复现结果：Pointpillar算法复现结果分析_kitti ap40 results-CSDN博客

参考博客：（三）PointPillars论文的MMDetection3D代码解读——数据处理篇_pointpillars代码-CSDN博客

一、简介

二、配置文件

（一）kitti-3d-3class.py

（二）pointpillars_hv_secfpn_kitti.py

（三）cyclic-40e.py

（四）default_runtime.py

三、kitti数据集处理

（一）kitti_dataset.py

（二）基类Det3DDataset的parse_data_info()函数

（三）KittiDataset的parse_ann_info函数

（四）基类Det3DDataset的parse_ann_info()函数

一、简介

PointPillars 是一个来自工业界的模型，整体的思想是基于图片的处理框架，直接将点云从俯视图的视角划分为一个个的立方柱体（Pillars），从而构成了伪图片数据，然后再使用2D检测框架进行特征提取和预测得到检测框，从而使得该模型在速度和精度都达到了一个很好的平衡。 PointPillars 的网络结构如图1.1所示。

二、配置文件

（一）kitti-3d-3class.py

代码存放于 mmdetection3d/config/_base_/datasets/kitti-3d-3class.py

代码全文如下：

# dataset settings
dataset_type = 'KittiDataset'
data_root = 'data/kitti/'
class_names = ['Pedestrian', 'Cyclist', 'Car']
point_cloud_range = [0, -40, -3, 70.4, 40, 1]
input_modality = dict(use_lidar=True, use_camera=False)
metainfo = dict(classes=class_names)

# Example to use different file client
# Method 1: simply set the data root and let the file I/O module
# automatically infer from prefix (not support LMDB and Memcache yet)

# data_root = 's3://openmmlab/datasets/detection3d/kitti/'

# Method 2: Use backend_args, file_client_args in versions before 1.1.0
# backend_args = dict(
#     backend='petrel',
#     path_mapping=dict({
#         './data/': 's3://openmmlab/datasets/detection3d/',
#          'data/': 's3://openmmlab/datasets/detection3d/'
#      }))
backend_args = None

db_sampler = dict(
    data_root=data_root,
    info_path=data_root + 'kitti_dbinfos_train.pkl',
    rate=1.0,
    prepare=dict(
        filter_by_difficulty=[-1],
        filter_by_min_points=dict(Car=5, Pedestrian=10, Cyclist=10)),
    classes=class_names,
    sample_groups=dict(Car=12, Pedestrian=6, Cyclist=6),
    points_loader=dict(
        type='LoadPointsFromFile',
        coord_type='LIDAR',
        load_dim=4,
        use_dim=4,
        backend_args=backend_args),
    backend_args=backend_args)

train_pipeline = [
    dict(
        type='LoadPointsFromFile',
        coord_type='LIDAR',
        load_dim=4,  # x, y, z, intensity
        use_dim=4,
        backend_args=backend_args),
    dict(type='LoadAnnotations3D', with_bbox_3d=True, with_label_3d=True),
    dict(type='ObjectSample', db_sampler=db_sampler),
    dict(
        type='ObjectNoise',
        num_try=100,
        translation_std=[1.0, 1.0, 0.5],
        global_rot_range=[0.0, 0.0],
        rot_range=[-0.78539816, 0.78539816]),
    dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),
    dict(
        type='GlobalRotScaleTrans',
        rot_range=[-0.78539816, 0.78539816],
        scale_ratio_range=[0.95, 1.05]),
    dict(type='PointsRangeFilter', point_cloud_range=point_cloud_range),
    dict(type='ObjectRangeFilter', point_cloud_range=point_cloud_range),
    dict(type='PointShuffle'),
    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,
        backend_args=backend_args),
    dict(
        type='MultiScaleFlipAug3D',
        img_scale=(1333, 800),
        pts_scale_ratio=1,
        flip=False,
        transforms=[
            dict(
                type='GlobalRotScaleTrans',
                rot_range=[0, 0],
                scale_ratio_range=[1., 1.],
                translation_std=[0, 0, 0]),
            dict(type='RandomFlip3D'),
            dict(
                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,
        backend_args=backend_args),
    dict(type='Pack3DDetInputs', keys=['points'])
]
train_dataloader = dict(
    batch_size=8,
    num_workers=4,
    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='kitti_infos_train.pkl',
            data_prefix=dict(pts='training/velodyne_reduced'),
            pipeline=train_pipeline,
            modality=input_modality,
            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',
            backend_args=backend_args)))
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,
        data_prefix=dict(pts='training/velodyne_reduced'),
        ann_file='kitti_infos_val.pkl',
        pipeline=test_pipeline,
        modality=input_modality,
        test_mode=True,
        metainfo=metainfo,
        box_type_3d='LiDAR',
        backend_args=backend_args))
test_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,
        data_prefix=dict(pts='training/velodyne_reduced'),
        ann_file='kitti_infos_val.pkl',
        pipeline=test_pipeline,
        modality=input_modality,
        test_mode=True,
        metainfo=metainfo,
        box_type_3d='LiDAR',
        backend_args=backend_args))
val_evaluator = dict(
    type='KittiMetric',
    ann_file=data_root + 'kitti_infos_val.pkl',
    metric='bbox',
    backend_args=backend_args)
test_evaluator = val_evaluator

vis_backends = [dict(type='LocalVisBackend')]
visualizer = dict(
    type='Det3DLocalVisualizer', vis_backends=vis_backends, name='visualizer')

以下是分部分解读：

# dataset settings
dataset_type = 'KittiDataset'
data_root = 'data/kitti/'
class_names = ['Pedestrian', 'Cyclist', 'Car']
point_cloud_range = [0, -40, -3, 70.4, 40, 1]
input_modality = dict(use_lidar=True, use_camera=False)
metainfo = dict(classes=class_names)

这里是kitti数据集的配置，其中数据类型是kitti的数据集（kittiDataset）、数据根目录在data/kitti、类别名称为“行人、骑行者、汽车”、第四行为点云范围、输入形式为激光雷达【未使用相机视觉】、metainfo是元信息，确定了类别名称。

db_sampler = dict(
    data_root=data_root,
    info_path=data_root + 'kitti_dbinfos_train.pkl',
    rate=1.0,
    prepare=dict(
        filter_by_difficulty=[-1],
        filter_by_min_points=dict(Car=5, Pedestrian=10, Cyclist=10)),
    classes=class_names,
    sample_groups=dict(Car=12, Pedestrian=6, Cyclist=6),
    points_loader=dict(
        type=LoadPointsFromFile,
        coord_type='LIDAR',
        load_dim=4,
        use_dim=4,
        backend_args=backend_args),
    backend_args=backend_args)

这里是kitti_dbinfos_train.pkl标注文件的读取，data_root定义数据存放的目录、info_path定义信息的路径。

train_dataloader = dict(
    batch_size=6,
    num_workers=4,
    persistent_workers=True,
    sampler=dict(type=DefaultSampler, shuffle=True),
    dataset=dict(
        type=RepeatDataset,
        times=2,
        dataset=dict(
            type=KittiDataset,
            data_root=data_root,
            ann_file='kitti_infos_train.pkl',
            data_prefix=dict(pts='training/velodyne_reduced'),
            pipeline=train_pipeline,
            modality=input_modality,
            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',
            backend_args=backend_args)))

在train_dataloader中定义了数据加载部分的信息，其中Batch_size（批处理大小）设置为6（即一次性处理6位数据），num_workers定义了工作区间的大小为4。

data_root定义数据集的路径、ann_file定义标注文件的路径、data_prefix定义点云文件的路径、pipeline这里用于定义train_pipeline负责读取点云文件和标注文件以及一些数据增强的操作。

train_pipeline定义如下：

train_pipeline = [
    dict(
        type=LoadPointsFromFile,
        coord_type='LIDAR',
        load_dim=4,  # x, y, z, intensity
        use_dim=4,
        backend_args=backend_args),
    dict(type=LoadAnnotations3D, with_bbox_3d=True, with_label_3d=True),
    dict(type=ObjectSample, db_sampler=db_sampler),
    dict(
        type=ObjectNoise,
        num_try=100,
        translation_std=[1.0, 1.0, 0.5],
        global_rot_range=[0.0, 0.0],
        rot_range=[-0.78539816, 0.78539816]),
    dict(type=RandomFlip3D, flip_ratio_bev_horizontal=0.5),
    dict(
        type=GlobalRotScaleTrans,
        rot_range=[-0.78539816, 0.78539816],
        scale_ratio_range=[0.95, 1.05]),
    dict(type=PointsRangeFilter, point_cloud_range=point_cloud_range),
    dict(type=ObjectRangeFilter, point_cloud_range=point_cloud_range),
    dict(type=PointShuffle),
    dict(
        type=Pack3DDetInputs, keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
]

第一个dict字典用于读取点云文件，第二个dict用于读取标注文件，第三到第九个dict用于点云数据的增强，最后一个dict将点云数据进行打包。

在val_dataloader和test_dataloader中的pipeline都以test_pipeline赋值，下面看test_pipeline定义：

test_pipeline = [
    dict(
        type=LoadPointsFromFile,
        coord_type='LIDAR',
        load_dim=4,
        use_dim=4,
        backend_args=backend_args),
    dict(
        type=MultiScaleFlipAug3D,
        img_scale=(1333, 800),
        pts_scale_ratio=1,
        flip=False,
        transforms=[
            dict(
                type=GlobalRotScaleTrans,
                rot_range=[0, 0],
                scale_ratio_range=[1., 1.],
                translation_std=[0, 0, 0]),
            dict(type=RandomFlip3D),
            dict(type=PointsRangeFilter, point_cloud_range=point_cloud_range)
        ]),
    dict(type=Pack3DDetInputs, keys=['points'])
]

一共三个字典定义，第一个dict用于读取点云文件，第二个dict用于点云数据的增强，第三个dict将点云数据进行打包。

（二）pointpillars_hv_secfpn_kitti.py

代码路径：mmdetection3d/config/_base_/models/pointpillars_hv_secfpn_kitti.py，全部代码：

voxel_size = [0.16, 0.16, 4]

model = dict(
    type='VoxelNet',
    data_preprocessor=dict(
        type='Det3DDataPreprocessor',
        voxel=True,
        voxel_layer=dict(
            max_num_points=32,  # max_points_per_voxel
            point_cloud_range=[0, -39.68, -3, 69.12, 39.68, 1],
            voxel_size=voxel_size,
            max_voxels=(16000, 40000))),
    voxel_encoder=dict(
        type='PillarFeatureNet',
        in_channels=4,

最低0.47元/天解锁文章

200万优质内容无限畅学