class EpochBasedTrainLoop in mmengine/runner/loops.py: class CocoDataset in mmdet/datasets/coco.py: need at least one array to concatenate
#9610
Comments
|
seems to be the same issue with #9613 |
|
Found out why. The metainfo has been changed to lowercase since #9469. On the custom dataset, use lowercase letters for keys in the dictionary instead. cfg.metainfo = {
'CLASSES': ('balloon', ),
'PALETTE': [
(220, 20, 60),
]
}to cfg.metainfo = {
'classes': ('balloon', ),
'palette': [
(220, 20, 60),
]
}should work. |
This works! Thanks a lot! |
|
怎么解决的 我一脸懵 |
|
Hi, the followings is the code stored in configs/detr/detr_r101_100e_coco.py, CODE: train_pipeline = [ optim_wrapper = dict( max_epochs = 150 param_scheduler = [ auto_scale_lr = dict(base_batch_size=16) |
|
@JunKaiLiao Did the solution above not work for you? |
|
@amanikiruga I found out that the class name in anntation file was wrong. After fixing that, it can train without any issue. Thank you!! |
|
I'm getting the same error, but I'm not using uppercase class names. Can anyone help? Traceback (most recent call last): During handling of the above exception, another exception occurred: Traceback (most recent call last): During handling of the above exception, another exception occurred: Traceback (most recent call last): here's my custom dataset class: @DATASETS.register_module() |
|
At the beginning, I corrected it according to the plan given above and it still didn’t work. “pip list“ to check your mmdet installation location；my mmdet is in anaconda3/....../site-packages/mmdet, change the file content in /site-packages/mmdet/coco.py. |
isn't this wrong? |
|
It's the same error that happened to me. Please, help me and welcome to any comments. My environment is in the docker desktop in win10 and I used the "Dockerfile" here and installed through it. config is like this below. # Inherit and overwrite part of the config based on this config
_base_ = './rtmdet_m_8xb32-300e_coco.py'
data_root = '/mmdetection/workspace2/dataset/excavator/' # dataset root
train_batch_size_per_gpu = 4
train_num_workers = 2
max_epochs = 10
stage2_num_epochs = 1
base_lr = 0.00008
metainfo = {
'classes': ('excavator', ),
'palette': [
(220, 20, 60),
]
}
train_dataloader = dict(
batch_size=train_batch_size_per_gpu,
num_workers=train_num_workers,
dataset=dict(
data_root=data_root,
metainfo=metainfo,
data_prefix=dict(img='train/'),
ann_file='train.json'))
val_dataloader = dict(
dataset=dict(
data_root=data_root,
metainfo=metainfo,
data_prefix=dict(img='val/'),
ann_file='val.json'))
test_dataloader = val_dataloader
val_evaluator = dict(ann_file=data_root + 'val.json')
test_evaluator = val_evaluator
model = dict(bbox_head=dict(num_classes=1))
# learning rate
param_scheduler = [
dict(
type='LinearLR',
start_factor=1.0e-5,
by_epoch=False,
begin=0,
end=10),
dict(
# use cosine lr from 10 to 20 epoch
type='CosineAnnealingLR',
eta_min=base_lr * 0.05,
begin=max_epochs // 2,
end=max_epochs,
T_max=max_epochs // 2,
by_epoch=True,
convert_to_iter_based=True),
]
train_pipeline_stage2 = [
dict(type='LoadImageFromFile', backend_args=None),
dict(type='LoadAnnotations', with_bbox=True),
dict(
type='RandomResize',
scale=(640, 640),
ratio_range=(0.1, 2.0),
keep_ratio=True),
dict(type='RandomCrop', crop_size=(640, 640)),
dict(type='YOLOXHSVRandomAug'),
dict(type='RandomFlip', prob=0.5),
dict(type='Pad', size=(640, 640), pad_val=dict(img=(114, 114, 114))),
dict(type='PackDetInputs')
]
# optimizer
optim_wrapper = dict(
_delete_=True,
type='OptimWrapper',
optimizer=dict(type='AdamW', lr=base_lr, weight_decay=0.05),
paramwise_cfg=dict(
norm_decay_mult=0, bias_decay_mult=0, bypass_duplicate=True))
default_hooks = dict(
checkpoint=dict(
interval=5,
max_keep_ckpts=2, # only keep latest 2 checkpoints
save_best='auto'
),
logger=dict(type='LoggerHook', interval=5))
custom_hooks = [
dict(
type='PipelineSwitchHook',
switch_epoch=max_epochs - stage2_num_epochs,
switch_pipeline=train_pipeline_stage2)
]
# load COCO pre-trained weight
load_from = '/mmdetection/checkpoints/rtmdet_m_8xb32-300e_coco_20220719_112220-229f527c.pth'
train_cfg = dict(type='EpochBasedTrainLoop', max_epochs=max_epochs, val_interval=1)
visualizer = dict(vis_backends=[dict(type='LocalVisBackend'),dict(type='TensorboardVisBackend')])Errros's here: Traceback (most recent call last):
File "/mmdetection/tools/train.py", line 133, in <module>
main()
File "/mmdetection/tools/train.py", line 129, in main
runner.train()
File "/opt/conda/lib/python3.9/site-packages/mmengine/runner/runner.py", line 1701, in train
self._train_loop = self.build_train_loop(
File "/opt/conda/lib/python3.9/site-packages/mmengine/runner/runner.py", line 1493, in build_train_loop
loop = LOOPS.build(
File "/opt/conda/lib/python3.9/site-packages/mmengine/registry/registry.py", line 570, in build
return self.build_func(cfg, *args, **kwargs, registry=self)
File "/opt/conda/lib/python3.9/site-packages/mmengine/registry/build_functions.py", line 144, in build_from_cfg
raise type(e)(
ValueError: class `EpochBasedTrainLoop` in mmengine/runner/loops.py: class `CocoDataset` in mmdet/datasets/coco.py: need at least one array to concatenate |
Prerequisite
Task
I'm using the official example scripts/configs for the officially supported tasks/models/datasets.
Branch
master branch https://github.com/open-mmlab/mmdetection
Environment
torch version: 1.11.0+cu113 cuda: True
mmdetection: 3.0.0rc5
mmcv: 2.0.0rc3
mmengine: 0.4.0
Reproduces the problem - code sample
start training
runner.train()
Reproduces the problem - command or script
start training
runner.train()
Reproduces the problem - error message
loading annotations into memory...
Done (t=0.01s)
creating index...
index created!
ValueError Traceback (most recent call last)
/opt/conda/lib/python3.7/site-packages/mmengine/registry/build_functions.py in build_from_cfg(cfg, registry, default_args)
120 else:
--> 121 obj = obj_cls(**args) # type: ignore
122
/kaggle/working/mmdetection/mmdet/datasets/base_det_dataset.py in init(self, seg_map_suffix, proposal_file, file_client_args, *args, **kwargs)
32 self.file_client = FileClient(**file_client_args)
---> 33 super().init(*args, **kwargs)
34
/opt/conda/lib/python3.7/site-packages/mmengine/dataset/base_dataset.py in init(self, ann_file, metainfo, data_root, data_prefix, filter_cfg, indices, serialize_data, pipeline, test_mode, lazy_init, max_refetch)
246 if not lazy_init:
--> 247 self.full_init()
248
/kaggle/working/mmdetection/mmdet/datasets/base_det_dataset.py in full_init(self)
70 if self.serialize_data:
---> 71 self.data_bytes, self.data_address = self._serialize_data()
72
/opt/conda/lib/python3.7/site-packages/mmengine/dataset/base_dataset.py in _serialize_data(self)
763 # TODO Check if np.concatenate is necessary
--> 764 data_bytes = np.concatenate(data_list)
765 # Empty cache for preventing making multiple copies of
<array_function internals> in concatenate(*args, **kwargs)
ValueError: need at least one array to concatenate
During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
/opt/conda/lib/python3.7/site-packages/mmengine/registry/build_functions.py in build_from_cfg(cfg, registry, default_args)
120 else:
--> 121 obj = obj_cls(**args) # type: ignore
122
/opt/conda/lib/python3.7/site-packages/mmengine/runner/loops.py in init(self, runner, dataloader, max_epochs, val_begin, val_interval, dynamic_intervals)
42 dynamic_intervals: Optional[List[Tuple[int, int]]] = None) -> None:
---> 43 super().init(runner, dataloader)
44 self._max_epochs = int(max_epochs)
/opt/conda/lib/python3.7/site-packages/mmengine/runner/base_loop.py in init(self, runner, dataloader)
26 self.dataloader = runner.build_dataloader(
---> 27 dataloader, seed=runner.seed, diff_rank_seed=diff_rank_seed)
28 else:
/opt/conda/lib/python3.7/site-packages/mmengine/runner/runner.py in build_dataloader(dataloader, seed, diff_rank_seed)
1332 if isinstance(dataset_cfg, dict):
-> 1333 dataset = DATASETS.build(dataset_cfg)
1334 if hasattr(dataset, 'full_init'):
/opt/conda/lib/python3.7/site-packages/mmengine/registry/registry.py in build(self, cfg, *args, **kwargs)
520 """
--> 521 return self.build_func(cfg, *args, **kwargs, registry=self)
522
/opt/conda/lib/python3.7/site-packages/mmengine/registry/build_functions.py in build_from_cfg(cfg, registry, default_args)
135 raise type(e)(
--> 136 f'class
{obj_cls.__name__}in ' # type: ignore137 f'{cls_location}.py: {e}')
ValueError: class
CocoDatasetin mmdet/datasets/coco.py: need at least one array to concatenateDuring handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
/tmp/ipykernel_23/3729266276.py in
1 # start training
----> 2 runner.train()
/opt/conda/lib/python3.7/site-packages/mmengine/runner/runner.py in train(self)
1647
1648 self._train_loop = self.build_train_loop(
-> 1649 self._train_loop) # type: ignore
1650
1651 #
build_optimizershould be called beforebuild_param_scheduler/opt/conda/lib/python3.7/site-packages/mmengine/runner/runner.py in build_train_loop(self, loop)
1441 loop_cfg,
1442 default_args=dict(
-> 1443 runner=self, dataloader=self._train_dataloader))
1444 else:
1445 by_epoch = loop_cfg.pop('by_epoch')
/opt/conda/lib/python3.7/site-packages/mmengine/registry/registry.py in build(self, cfg, *args, **kwargs)
519 >>> model = MODELS.build(cfg)
520 """
--> 521 return self.build_func(cfg, *args, **kwargs, registry=self)
522
523 def _add_child(self, registry: 'Registry') -> None:
/opt/conda/lib/python3.7/site-packages/mmengine/registry/build_functions.py in build_from_cfg(cfg, registry, default_args)
134 obj_cls.module.split('.')) # type: ignore
135 raise type(e)(
--> 136 f'class
{obj_cls.__name__}in ' # type: ignore137 f'{cls_location}.py: {e}')
138
ValueError: class
EpochBasedTrainLoopin mmengine/runner/loops.py: classCocoDatasetin mmdet/datasets/coco.py: need at least one array to concatenateAdditional information
I ran this tutorial (https://github.com/open-mmlab/mmdetection/blob/3.x/demo/MMDet_InstanceSeg_Tutorial.ipynb) without changing any code, but I ran into a problem when I get to runner.train(), how should I fix it?
The text was updated successfully, but these errors were encountered: