search

open-mmlab / mmdetection3d

OpenMMLab's next-generation platform for general 3D object detection.
https://mmdetection3d.readthedocs.io/en/latest/
Apache License 2.0
5.36k stars 1.55k forks source link

[Bug] 3DSSD not training when batch size greater 1 #2059

Closed holtvogt closed 1 year ago

holtvogt commented 2 years ago

Prerequisite

Task

I have modified the scripts/configs, or I'm working on my own tasks/models/datasets.

Branch

master branch https://github.com/open-mmlab/mmdetection3d

Environment

sys.platform: linux
Python: 3.8.15 (default, Nov  4 2022, 20:59:55) [GCC 11.2.0]
CUDA available: True
GPU 0: NVIDIA GeForce GTX 1080 Ti
CUDA_HOME: /usr/local/cuda
NVCC: Cuda compilation tools, release 11.6, V11.6.124
GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0
PyTorch: 1.10.1
PyTorch compiling details: PyTorch built with:
  - GCC 7.3
  - C++ Version: 201402
  - Intel(R) oneAPI Math Kernel Library Version 2021.4-Product Build 20210904 for Intel(R) 64 architecture applications
  - Intel(R) MKL-DNN v2.2.3 (Git Hash 7336ca9f055cf1bfa13efb658fe15dc9b41f0740)
  - OpenMP 201511 (a.k.a. OpenMP 4.5)
  - LAPACK is enabled (usually provided by MKL)
  - NNPACK is enabled
  - CPU capability usage: AVX2
  - CUDA Runtime 11.3
  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_80,code=sm_80;-gencode;arch=compute_86,code=sm_86;-gencode;arch=compute_37,code=compute_37
  - CuDNN 8.2
  - Magma 2.5.2
  - Build settings: BLAS_INFO=mkl, BUILD_TYPE=Release, CUDA_VERSION=11.3, CUDNN_VERSION=8.2.0, CXX_COMPILER=/opt/rh/devtoolset-7/root/usr/bin/c++, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_KINETO -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -DEDGE_PROFILER_USE_KINETO -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-unused-local-typedefs -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-psabi -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, LAPACK_INFO=mkl, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, TORCH_VERSION=1.10.1, USE_CUDA=ON, USE_CUDNN=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, 

TorchVision: 0.11.2
OpenCV: 4.6.0
MMCV: 1.6.2
MMCV Compiler: GCC 9.3
MMCV CUDA Compiler: 11.3
MMDetection: 2.26.0
MMSegmentation: 0.29.1
MMDetection3D: 1.0.0rc4+482141b
spconv2.0: True

Reproduces the problem - code sample

My 3DSSD configuration:

model = dict(
    type='SSD3DNet',
    backbone=dict(
        type='PointNet2SAMSG',
        in_channels=4,
        num_points=(4096, 512, (256, 256)),
        radii=((0.2, 0.4, 0.8), (0.4, 0.8, 1.6), (1.6, 3.2, 4.8)),
        num_samples=((32, 32, 64), (32, 32, 64), (32, 32, 32)),
        sa_channels=(((16, 16, 32), (16, 16, 32), (32, 32, 64)),
                     ((64, 64, 128), (64, 64, 128), (64, 96, 128)),
                     ((128, 128, 256), (128, 192, 256), (128, 256, 256))),
        aggregation_channels=(64, 128, 256),
        fps_mods=('D-FPS', 'FS', ('F-FPS', 'D-FPS')),
        fps_sample_range_lists=(-1, -1, (512, -1)),
        norm_cfg=dict(type='BN2d', eps=0.001, momentum=0.1),
        sa_cfg=dict(
            type='PointSAModuleMSG',
            pool_mod='max',
            use_xyz=True,
            normalize_xyz=False)),
    bbox_head=dict(
        type='SSD3DHead',
        in_channels=256,
        vote_module_cfg=dict(
            in_channels=256,
            num_points=256,
            gt_per_seed=1,
            conv_channels=(128, ),
            conv_cfg=dict(type='Conv1d'),
            norm_cfg=dict(type='BN1d', eps=0.001, momentum=0.1),
            with_res_feat=False,
            vote_xyz_range=(3.0, 3.0, 2.0)),
        vote_aggregation_cfg=dict(
            type='PointSAModuleMSG',
            num_point=256,
            radii=(4.8, 6.4),
            sample_nums=(16, 32),
            mlp_channels=((256, 256, 256, 512), (256, 256, 512, 1024)),
            norm_cfg=dict(type='BN2d', eps=0.001, momentum=0.1),
            use_xyz=True,
            normalize_xyz=False,
            bias=True),
        pred_layer_cfg=dict(
            in_channels=1536,
            shared_conv_channels=(512, 128),
            cls_conv_channels=(128, ),
            reg_conv_channels=(128, ),
            conv_cfg=dict(type='Conv1d'),
            norm_cfg=dict(type='BN1d', eps=0.001, momentum=0.1),
            bias=True),
        conv_cfg=dict(type='Conv1d'),
        norm_cfg=dict(type='BN1d', eps=0.001, momentum=0.1),
        objectness_loss=dict(
            type='CrossEntropyLoss',
            use_sigmoid=True,
            reduction='sum',
            loss_weight=1.0),
        center_loss=dict(
            type='SmoothL1Loss', reduction='sum', loss_weight=1.0),
        dir_class_loss=dict(
            type='CrossEntropyLoss', reduction='sum', loss_weight=1.0),
        dir_res_loss=dict(
            type='SmoothL1Loss', reduction='sum', loss_weight=1.0),
        size_res_loss=dict(
            type='SmoothL1Loss', reduction='sum', loss_weight=1.0),
        corner_loss=dict(
            type='SmoothL1Loss', reduction='sum', loss_weight=1.0),
        vote_loss=dict(type='SmoothL1Loss', reduction='sum', loss_weight=1.0),
        num_classes=1,
        bbox_coder=dict(
            type='AnchorFreeBBoxCoder', num_dir_bins=12, with_rot=True)),
    train_cfg=dict(
        sample_mod='spec', pos_distance_thr=10.0, expand_dims_length=0.05),
    test_cfg=dict(
        nms_cfg=dict(type='nms', iou_thr=0.1),
        sample_mod='spec',
        score_thr=0.0,
        per_class_proposal=True,
        max_output_num=100))
dataset_type = 'ItivDataset'
data_root = 'data/itiv/'
class_names = ['Pedestrian']
point_cloud_range = [-1, -3, -2.5, 3, 0.5, 1]
input_modality = dict(use_lidar=True, use_camera=False)
file_client_args = dict(backend='disk')
db_sampler = dict(
    data_root='data/itiv/',
    info_path='data/itiv/itiv_dbinfos_train.pkl',
    rate=1.0,
    prepare=dict(
        filter_by_difficulty=[-1], filter_by_min_points=dict(Pedestrian=5)),
    classes=['Pedestrian'],
    sample_groups=dict(Pedestrian=15))
train_pipeline = [
    dict(
        type='LoadPointsFromFile',
        coord_type='LIDAR',
        load_dim=4,
        use_dim=4,
        file_client_args=dict(backend='disk')),
    dict(
        type='LoadAnnotations3D',
        with_bbox_3d=True,
        with_label_3d=True,
        file_client_args=dict(backend='disk')),
    dict(
        type='ObjectSample',
        db_sampler=dict(
            data_root='data/itiv/',
            info_path='data/itiv/itiv_dbinfos_train.pkl',
            rate=1.0,
            prepare=dict(
                filter_by_difficulty=[-1],
                filter_by_min_points=dict(Pedestrian=5)),
            classes=['Pedestrian'],
            sample_groups=dict(Pedestrian=15))),
    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=[-1, -3, -2.5, 3, 0.5, 1]),
    dict(
        type='ObjectRangeFilter', point_cloud_range=[-1, -3, -2.5, 3, 0.5, 1]),
    dict(type='PointShuffle'),
    dict(type='DefaultFormatBundle3D', class_names=['Pedestrian']),
    dict(type='Collect3D', keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
]
test_pipeline = [
    dict(
        type='LoadPointsFromFile',
        coord_type='LIDAR',
        load_dim=4,
        use_dim=4,
        file_client_args=dict(backend='disk')),
    dict(
        type='MultiScaleFlipAug3D',
        img_scale=(1333, 800),
        pts_scale_ratio=1,
        flip=False,
        transforms=[
            dict(
                type='GlobalRotScaleTrans',
                rot_range=[0, 0],
                scale_ratio_range=[1.0, 1.0],
                translation_std=[0, 0, 0]),
            dict(type='RandomFlip3D'),
            dict(
                type='PointsRangeFilter',
                point_cloud_range=[-1, -3, -2.5, 3, 0.5, 1]),
            dict(
                type='DefaultFormatBundle3D',
                class_names=['Pedestrian'],
                with_label=False),
            dict(type='Collect3D', keys=['points'])
        ])
]
eval_pipeline = [
    dict(
        type='LoadPointsFromFile',
        coord_type='LIDAR',
        load_dim=4,
        use_dim=4,
        file_client_args=dict(backend='disk')),
    dict(
        type='DefaultFormatBundle3D',
        class_names=['Pedestrian'],
        with_label=False),
    dict(type='Collect3D', keys=['points'])
]
data = dict(
    samples_per_gpu=16,
    workers_per_gpu=4,
    train=dict(
        type='RepeatDataset',
        times=1,
        dataset=dict(
            type='ItivDataset',
            data_root='data/itiv/',
            ann_file='data/itiv/itiv_infos_train.pkl',
            split='training',
            pts_prefix='velodyne',
            pipeline=[
                dict(
                    type='LoadPointsFromFile',
                    coord_type='LIDAR',
                    load_dim=4,
                    use_dim=4,
                    file_client_args=dict(backend='disk')),
                dict(
                    type='LoadAnnotations3D',
                    with_bbox_3d=True,
                    with_label_3d=True,
                    file_client_args=dict(backend='disk')),
                dict(
                    type='ObjectSample',
                    db_sampler=dict(
                        data_root='data/itiv/',
                        info_path='data/itiv/itiv_dbinfos_train.pkl',
                        rate=1.0,
                        prepare=dict(
                            filter_by_difficulty=[-1],
                            filter_by_min_points=dict(Pedestrian=5)),
                        classes=['Pedestrian'],
                        sample_groups=dict(Pedestrian=15))),
                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=[-1, -3, -2.5, 3, 0.5, 1]),
                dict(
                    type='ObjectRangeFilter',
                    point_cloud_range=[-1, -3, -2.5, 3, 0.5, 1]),
                dict(type='PointShuffle'),
                dict(type='DefaultFormatBundle3D', class_names=['Pedestrian']),
                dict(
                    type='Collect3D',
                    keys=['points', 'gt_bboxes_3d', 'gt_labels_3d'])
            ],
            modality=dict(use_lidar=True, use_camera=False),
            classes=['Pedestrian'],
            test_mode=False,
            box_type_3d='LiDAR')),
    val=dict(
        type='ItivDataset',
        data_root='data/itiv/',
        ann_file='data/itiv/itiv_infos_val.pkl',
        split='training',
        pts_prefix='velodyne',
        pipeline=[
            dict(
                type='LoadPointsFromFile',
                coord_type='LIDAR',
                load_dim=4,
                use_dim=4,
                file_client_args=dict(backend='disk')),
            dict(
                type='MultiScaleFlipAug3D',
                img_scale=(1333, 800),
                pts_scale_ratio=1,
                flip=False,
                transforms=[
                    dict(
                        type='GlobalRotScaleTrans',
                        rot_range=[0, 0],
                        scale_ratio_range=[1.0, 1.0],
                        translation_std=[0, 0, 0]),
                    dict(type='RandomFlip3D'),
                    dict(
                        type='PointsRangeFilter',
                        point_cloud_range=[-1, -3, -2.5, 3, 0.5, 1]),
                    dict(
                        type='DefaultFormatBundle3D',
                        class_names=['Pedestrian'],
                        with_label=False),
                    dict(type='Collect3D', keys=['points'])
                ])
        ],
        modality=dict(use_lidar=True, use_camera=False),
        classes=['Pedestrian'],
        test_mode=True,
        box_type_3d='LiDAR'),
    test=dict(
        type='ItivDataset',
        data_root='data/itiv/',
        ann_file='data/itiv/itiv_infos_val.pkl',
        split='training',
        pts_prefix='velodyne',
        pipeline=[
            dict(
                type='LoadPointsFromFile',
                coord_type='LIDAR',
                load_dim=4,
                use_dim=4,
                file_client_args=dict(backend='disk')),
            dict(
                type='MultiScaleFlipAug3D',
                img_scale=(1333, 800),
                pts_scale_ratio=1,
                flip=False,
                transforms=[
                    dict(
                        type='GlobalRotScaleTrans',
                        rot_range=[0, 0],
                        scale_ratio_range=[1.0, 1.0],
                        translation_std=[0, 0, 0]),
                    dict(type='RandomFlip3D'),
                    dict(
                        type='PointsRangeFilter',
                        point_cloud_range=[-1, -3, -2.5, 3, 0.5, 1]),
                    dict(
                        type='DefaultFormatBundle3D',
                        class_names=['Pedestrian'],
                        with_label=False),
                    dict(type='Collect3D', keys=['points'])
                ])
        ],
        modality=dict(use_lidar=True, use_camera=False),
        classes=['Pedestrian'],
        test_mode=True,
        box_type_3d='LiDAR'))
evaluation = dict(
    interval=100000,
    pipeline=[
        dict(
            type='LoadPointsFromFile',
            coord_type='LIDAR',
            load_dim=4,
            use_dim=4,
            file_client_args=dict(backend='disk')),
        dict(
            type='DefaultFormatBundle3D',
            class_names=['Pedestrian'],
            with_label=False),
        dict(type='Collect3D', keys=['points'])
    ])
checkpoint_config = dict(interval=10)
log_config = dict(
    interval=10,
    hooks=[
        dict(type='TextLoggerHook'),
        dict(type='CustomTensorboardLoggerHook')
    ])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = '/home/ws/x/Dokumente/Repositories/dasad3d/checkpoints/itiv/3dssd/batchsize16_epochs50'
load_from = None
resume_from = None
workflow = [('train', 1), ('val', 1)]
opencv_num_threads = 0
mp_start_method = 'fork'
lr = 0.002
optimizer = dict(type='AdamW', lr=0.002, weight_decay=0)
optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
lr_config = dict(policy='step', warmup=None, step=[45, 60])
runner = dict(type='EpochBasedRunner', max_epochs=50)
gpu_ids = [0]

Reproduces the problem - error message

Traceback (most recent call last):
  File "./tools/train.py", line 298, in <module>
    main()
  File "./tools/train.py", line 286, in main
    train_model(
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/apis/train.py", line 407, in train_model
    train_detector(
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/apis/train.py", line 379, in train_detector
    runner.run(data_loaders, cfg.workflow)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 136, in run
    epoch_runner(data_loaders[i], **kwargs)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 53, in train
    self.run_iter(data_batch, train_mode=True, **kwargs)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 31, in run_iter
    outputs = self.model.train_step(data_batch, self.optimizer,
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/parallel/data_parallel.py", line 77, in train_step
    return self.module.train_step(*inputs[0], **kwargs[0])
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmdet/models/detectors/base.py", line 248, in train_step
    losses = self(**data)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/runner/fp16_utils.py", line 116, in new_func
    return old_func(*args, **kwargs)
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/models/detectors/base.py", line 63, in forward
    return self.forward_train(**kwargs)
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/models/detectors/votenet.py", line 58, in forward_train
    points_cat = torch.stack(points)
RuntimeError: stack expects each tensor to be equal size, but got [36220, 4] at entry 0 and [36804, 4] at entry 1

Additional information

The training stops immediately with the abovementioned error message as soon as I increase the batch size to greater than 1. Any ideas about what goes wrong?

JingweiZhang12 commented 2 years ago

Did you remove the PointSample in the training pipeline? It would cause the different numbers of points in the different samples.

holtvogt commented 2 years ago

Great note, I indeed missed including it. But after adding, I receive the following issue:

Traceback (most recent call last):
  File "./tools/train.py", line 298, in <module>
    main()
  File "./tools/train.py", line 286, in main
    train_model(
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/apis/train.py", line 407, in train_model
    train_detector(
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/apis/train.py", line 379, in train_detector
    runner.run(data_loaders, cfg.workflow)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 136, in run
    epoch_runner(data_loaders[i], **kwargs)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 49, in train
    for i, data_batch in enumerate(self.data_loader):
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 521, in __next__
    data = self._next_data()
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 1203, in _next_data
    return self._process_data(data)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 1229, in _process_data
    data.reraise()
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/_utils.py", line 434, in reraise
    raise exception
AttributeError: Caught AttributeError in DataLoader worker process 0.
Original Traceback (most recent call last):
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/utils/data/_utils/worker.py", line 287, in _worker_loop
    data = fetcher.fetch(index)
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 49, in fetch
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 49, in <listcomp>
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "/home/ws/x/anaconda3/envs/das3d/lib/python3.8/site-packages/mmdet/datasets/dataset_wrappers.py", line 178, in __getitem__
    return self.dataset[idx % self._ori_len]
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/datasets/custom_3d.py", line 449, in __getitem__
    data = self.prepare_train_data(idx)
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/datasets/custom_3d.py", line 236, in prepare_train_data
    example = self.pipeline(input_dict)
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/datasets/pipelines/compose.py", line 49, in __call__
    data = t(data)
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/datasets/pipelines/transforms_3d.py", line 1166, in __call__
    points, choices = self._points_random_sampling(
  File "/home/ws/x/Dokumente/Repositories/dasad3d/mmdet3d/datasets/pipelines/transforms_3d.py", line 1132, in _points_random_sampling
    replace = points.shape[0] < num_samples
AttributeError: 'DataContainer' object has no attribute 'shape'
JingweiZhang12 commented 1 year ago

@holtvogt . Maybe the position of PointSample you added is wrong. Please refer to this:https://github.com/open-mmlab/mmdetection3d/blob/master/configs/point_rcnn/point_rcnn_2x8_kitti-3d-3classes.py#L41