1. Introduction
MMdetection的特点：
- 模块化设计：将不同网络的部分进行切割，模块之间具有很高的复用性和独立性（十分便利，可以任意组合）
- 高效的内存使用
- 支持多种框架
- SOTA
2. Support Frameworks
-
单阶段检测器
SSD(2015)、RetinaNet(2017)、GHM(2019)、FCOS(2019)、FSAF(2019)
-
两阶段检测器
Faster R-CNN(2015)、R-FCN(2016)、Mask R-CNN(2017)、Grid R-CNN(2018)、Mask Scoring R-CNN(2019)、Double-Head R-CNN(2019)
-
多阶段检测器
Cascade R-CNN(2017)、Hybrid Task Cascade(2019)
-
通用模块和方法
- soft-NMS(2017)、DCN(2017)、OHEN(2016)、DCN2(2018)、Train from Scratch(2018)、ScratchDet(2018)、M2Det (2018)、GCNet(2019) 、Generalized Attention(2019)、SyncBN(2018)、Group Normalization(2018)、Weight Standardization(2019)、HRNet(2019) 、Guided Anchoring(2019)、Libra R-CNN(2019)
3. Architecture
模型表征：划分为以下几个模块：
Backbone（ResNet等）、Neck（FPN）、DenseHead（AnchorHead）、RoIExtractor、RoIHead（BBoxHead/MaskHead）
结构图如下：
4. Benchmarks
-
Datasets
支持COCO-sytle和 VOC-style数据集
-
Implementation details
Images are resized to a maximum scale of 1333 × 800,without changing the aspect ratio.
“1x” and “2x” means 12 epochs and 24 epochs respectively. “20e” is adopted in cascade models, which denotes 20 epochs.
-
Evaluation metrics
Adopting standard evaluation metrics for COCO dataset, where multiple IoU thresholds from 0.5 to 0.95 are applied.
The results of region proposal network (RPN) are measured with Average Recall (AR) and detection results are evaluated with mAP.
5. Config文件说明
model
backbone：通常是FCN，用于提取特征图，例如ResNet
neck：
- rpn_head
- box_roi_extractor：
- bbox_head
train_cfg
test_cfg
数据管道
train_pipeline
test_pipeline
# model settings
model = dict(
type='FastRCNN',
pretrained='torchvision://resnet50',
backbone=dict(
type='ResNet',
depth=50,
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),
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=81,
target_means=[0., 0., 0., 0.],
target_stds=[0.1, 0.1, 0.2, 0.2],
reg_class_agnostic=False,
loss_cls=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
loss_bbox=dict(type='SmoothL1Loss', beta=1.0, loss_weight=1.0)))
# model training and testing settings
train_cfg = dict(
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='RandomSampler',
num=512,
pos_fraction=0.25,
neg_pos_ub=-1,
add_gt_as_proposals=True),
pos_weight=-1,
debug=False))
test_cfg = dict(
rcnn=dict(
score_thr=0.05, nms=dict(type='nms', iou_thr=0.5), max_per_img=100))
# dataset settings
dataset_type = 'CocoDataset'
data_root = 'data/coco/'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
train_pipeline = [
# 数据加载
dict(type='LoadImageFromFile'),
dict(type='LoadProposals', num_max_proposals=2000),
dict(type='LoadAnnotations', with_bbox=True),
# 预处理
dict(type='Resize', img_scale=(1333, 800), keep_ratio=True),
dict(type='RandomFlip', flip_ratio=0.5),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size_divisor=32),
# 格式化
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'proposals', 'gt_bboxes', 'gt_labels']),
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadProposals', num_max_proposals=None),
dict(
type='MultiScaleFlipAug',
img_scale=(1333, 800),
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size_divisor=32),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img', 'proposals']),
])
]
data = dict(
imgs_per_gpu=2,
workers_per_gpu=2,
train=dict(
type=dataset_type,
ann_file=data_root + 'annotations/instances_train2017.json',
proposal_file=data_root + 'proposals/rpn_r50_fpn_1x_train2017.pkl',
img_prefix=data_root + 'train2017/',
pipeline=train_pipeline),
val=dict(
type=dataset_type,
ann_file=data_root + 'annotations/instances_val2017.json',
proposal_file=data_root + 'proposals/rpn_r50_fpn_1x_val2017.pkl',
img_prefix=data_root + 'val2017/',
pipeline=test_pipeline),
test=dict(
type=dataset_type,
ann_file=data_root + 'annotations/instances_val2017.json',
proposal_file=data_root + 'proposals/rpn_r50_fpn_1x_val2017.pkl',
img_prefix=data_root + 'val2017/',
pipeline=test_pipeline))
# optimizer
optimizer = dict(type='SGD', lr=0.02, 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=1)
# yapf:disable
log_config = dict(
interval=50,
hooks=[
dict(type='TextLoggerHook'),
# dict(type='TensorboardLoggerHook')
])
# yapf:enable
# runtime settings
total_epochs = 12
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/fast_rcnn_r50_fpn_1x'
load_from = None
resume_from = None
workflow = [('train', 1)]
