Mmdetection: Cascade R-CNN with OHEM

[ZengyuanYu]

@hellock Thanks your work! I use mmdetection for my finally thesis. Now I meet some question:

• 
  
  1. I want to compare OHEM and Random use Cascade R-CNN, But got error. I don't debug it, So will you give me some advice?
     
     my config
  
  

model settings

model = dict(
type='CascadeRCNN',
num_stages=3,
pretrained='modelzoo://resnet101',
backbone=dict(
type='ResNet',
depth=101,
num_stages=4,
out_indices=(0, 1, 2, 3),
frozen_stages=1,
style='pytorch'),
neck=dict(
type='FPN',
in_channels=[256, 512, 1024, 2048],
out_channels=256,
num_outs=5),
rpn_head=dict(
type='RPNHead',
in_channels=256,
feat_channels=256,
anchor_scales=[8],
anchor_ratios=[0.5, 1.0, 2.0],
anchor_strides=[4, 8, 16, 32, 64],
target_means=[.0, .0, .0, .0],
target_stds=[1.0, 1.0, 1.0, 1.0],
use_sigmoid_cls=True),
bbox_roi_extractor=dict(
type='SingleRoIExtractor',
roi_layer=dict(type='RoIAlign', out_size=7, sample_num=2),
out_channels=256,
featmap_strides=[4, 8, 16, 32]),
bbox_head=[
dict(
type='SharedFCBBoxHead',
num_fcs=2,
in_channels=256,
fc_out_channels=1024,
roi_feat_size=7,
num_classes=2,
target_means=[0., 0., 0., 0.],
target_stds=[0.1, 0.1, 0.2, 0.2],
reg_class_agnostic=True),
dict(
type='SharedFCBBoxHead',
num_fcs=2,
in_channels=256,
fc_out_channels=1024,
roi_feat_size=7,
num_classes=2,
target_means=[0., 0., 0., 0.],
target_stds=[0.05, 0.05, 0.1, 0.1],
reg_class_agnostic=True),
dict(
type='SharedFCBBoxHead',
num_fcs=2,
in_channels=256,
fc_out_channels=1024,
roi_feat_size=7,
num_classes=2,
target_means=[0., 0., 0., 0.],
target_stds=[0.033, 0.033, 0.067, 0.067],
reg_class_agnostic=True)
])

model training and testing settings

train_cfg = dict(
rpn=dict(
assigner=dict(
type='MaxIoUAssigner',
pos_iou_thr=0.7,
neg_iou_thr=0.3,
min_pos_iou=0.3,
ignore_iof_thr=-1),
sampler=dict(
type='RandomSampler',
num=256,
pos_fraction=0.5,
neg_pos_ub=-1,
add_gt_as_proposals=False),
allowed_border=0,
pos_weight=-1,
smoothl1_beta=1 / 9.0,
debug=False),
rcnn=[
dict(
assigner=dict(
type='MaxIoUAssigner',
pos_iou_thr=0.5,
neg_iou_thr=0.5,
min_pos_iou=0.5,
ignore_iof_thr=-1),
sampler=dict(
type='OHEMSampler',
num=512,
pos_fraction=0.25,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=-1,
debug=False),
dict(
assigner=dict(
type='MaxIoUAssigner',
pos_iou_thr=0.6,
neg_iou_thr=0.6,
min_pos_iou=0.6,
ignore_iof_thr=-1),
sampler=dict(
type='OHEMSampler',
num=512,
pos_fraction=0.25,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=-1,
debug=False),
dict(
assigner=dict(
type='MaxIoUAssigner',
pos_iou_thr=0.7,
neg_iou_thr=0.7,
min_pos_iou=0.7,
ignore_iof_thr=-1),
sampler=dict(
type='OHEMSampler',
num=512,
pos_fraction=0.25,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=-1,
debug=False)
],
stage_loss_weights=[1, 0.5, 0.25])
test_cfg = dict(
rpn=dict(
nms_across_levels=False,
nms_pre=2000,
nms_post=2000,
max_num=2000,
nms_thr=0.7,
min_bbox_size=0),
rcnn=dict(
score_thr=0.05, nms=dict(type='nms', iou_thr=0.5), max_per_img=100),
keep_all_stages=False)

dataset settings

dataset_type = 'CocoDataset'
data_root = '/home/yu/mmdetection/data/coco-wheat/'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
data = dict(
imgs_per_gpu=1,
workers_per_gpu=1,
train=dict(
type=dataset_type,
ann_file=data_root + 'annotations/trainval.json',
img_prefix=data_root + 'trainval',
img_scale=(1333, 800),
img_norm_cfg=img_norm_cfg,
size_divisor=32,
flip_ratio=0.5,
with_mask=False,
with_crowd=True,
with_label=True),
val=dict(
type=dataset_type,
ann_file=data_root + 'annotations/val.json',
img_prefix=data_root + 'val/',
img_scale=(1333, 800),
img_norm_cfg=img_norm_cfg,
size_divisor=32,
flip_ratio=0,
with_mask=False,
with_crowd=True,
with_label=True),
test=dict(
type=dataset_type,
ann_file=data_root + 'annotations/test.json',
img_prefix=data_root + 'test/',
img_scale=(1333, 800),
img_norm_cfg=img_norm_cfg,
size_divisor=32,
flip_ratio=0,
with_mask=False,
with_label=False,
test_mode=True))

optimizer

optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
optimizer_config = dict(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])
checkpoint_config = dict(interval=10)

yapf:disable

log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
# dict(type='TensorboardLoggerHook')
])

yapf:enable

runtime settings

total_epochs = 200
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/cascade_rcnn_r101_fpn_1x'
load_from = None
resume_from = None
workflow = [('train', 1)]

error msg
loading annotations into memory...
Done (t=0.08s)
creating index...
index created!
2019-02-24 20:31:17,671 - INFO - Start running, host: yu@xc-pc, work_dir: /home/yu/mmdetection/data/result/cascade_rcnn_r101_OHEM_wheat
2019-02-24 20:31:17,671 - INFO - workflow: [('train', 1)], max: 200 epochs
Traceback (most recent call last):
File "tools/train.py", line 90, in
main()
File "tools/train.py", line 86, in main
logger=logger)
File "/home/yu/mmdetection/mmdet/apis/train.py", line 59, in train_detector
_non_dist_train(model, dataset, cfg, validate=validate)
File "/home/yu/mmdetection/mmdet/apis/train.py", line 121, in _non_dist_train
runner.run(data_loaders, cfg.workflow, cfg.total_epochs)
File "/home/yu/.virtualenvs/mmdetetion/lib/python3.6/site-packages/mmcv/runner/runner.py", line 349, in run
epoch_runner(data_loaders[i], kwargs)
File "/home/yu/.virtualenvs/mmdetetion/lib/python3.6/site-packages/mmcv/runner/runner.py", line 255, in train
self.model, data_batch, train_mode=True, *kwargs)
File "/home/yu/mmdetection/mmdet/apis/train.py", line 37, in batch_processor
losses = model(data)
File "/home/yu/.virtualenvs/mmdetetion/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in __call__
result = self.forward(input, kwargs)
File "/home/yu/.virtualenvs/mmdetetion/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 121, in forward
return self.module(inputs[0], *kwargs[0])
File "/home/yu/.virtualenvs/mmdetetion/lib/python3.6/site-packages/torch/nn/modules/module.py", line 477, in __call__
result = self.forward(input, *kwargs)
File "/home/yu/mmdetection/mmdet/models/detectors/base.py", line 80, in forward
return self.forward_train(img, img_meta, *kwargs)
File "/home/yu/mmdetection/mmdet/models/detectors/cascade_rcnn.py", line 141, in forward_train
cfg=rcnn_train_cfg)
File "/home/yu/mmdetection/mmdet/core/utils/misc.py", line 24, in multi_apply
return tuple(map(list, zip(map_results)))
File "/home/yu/mmdetection/mmdet/core/bbox/assign_sampling.py", line 30, in assign_and_sample
bbox_sampler = build_sampler(cfg.sampler)
File "/home/yu/mmdetection/mmdet/core/bbox/assign_sampling.py", line 22, in build_sampler
cfg, samplers, default_args=kwargs)
File "/home/yu/.virtualenvs/mmdetetion/lib/python3.6/site-packages/mmcv/runner/utils.py", line 72, in obj_from_dict
return obj_type(args)
TypeError: __init__() missing 1 required positional argument: 'context'

[ZengyuanYu]

I use Faster R-CNN with OHEM is OK.

[hellock]

We have not tried OHEM in Cascade R-CNN and it is currently not supported. You can refer to here and here if you want to add OHEM in Cascade R-CNN.

[ZengyuanYu]

We have not tried OHEM in Cascade R-CNN and it is currently not supported. You can refer to here and here if you want to add OHEM in Cascade R-CNN.

OK, I will do it for refer code.Thank you!

[1224wxwx]

I use Faster R-CNN with OHEM is OK.

Could you please tell me whether OHEM helped you?
After I used OHEM, it became very difficult for loss to decrease(both on train dataset and test dataset) and the result was worse than before.

[zvadaszi]

Did OHEM helped with Cascade R-CNN?