mmdetection环境配置，三个目标检测模型训练

一、colab环境配置

依次执行以下命令，在这里我用的是VOC数据集

cd /content/drive/MyDrive

!pip install torch==1.5.0 torchvision==0.6.0

!pip install mmcv-full==latest+torch1.5.0+cu101 -f https://openmmlab.oss-accelerate.aliyuncs.com/mmcv/dist/index.html

!git clone https://github.com.cnpmjs.org/open-mmlab/mmdetection.git

cd /content/drive/MyDrive/mmdetection

!pip install -r requirements/build.txt

!pip install -v -e . # or "python setup.py develop"

!python tools/train.py -h

cd /content/drive/MyDrive/mmdetection

!mkdir data

!mkdir data/VOCdevkit

cd /content/drive/MyDrive/mmdetection

!python tools/train.py ./configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py
#!python tools/train.py ./configs/retinanet/retinanet_r50_fpn_1x_coco.py
#!python tools/train.py ./configs/cascade_rcnn/cascade_rcnn_r50_fpn_1x_coco.py

二、修改mmdetection中的代码

注意：大部分含有逗号的地方，逗号都是需要保存的

1、数据集

所有的数据标签存放在：./data/VOCdevkit/VOC2007/Annotations
所有的图片数据存放在：./data/VOCdevkit/VOC2007/JPEGImage

2、修改configs

修改：./mmdetection/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py
将其中的coco_detection修改为voc0712

_base_ = [
    '../_base_/models/faster_rcnn_r50_fpn.py',
    '../_base_/datasets/voc0712.py',
    '../_base_/schedules/schedule_1x.py', '../_base_/default_runtime.py'
]

其他重要文件：
…/base/models/faster_rcnn_r50_fpn.py：定义模型文件
…/base/datasets/coco_detection.py：定义训练数据路径等
…/base/schedules/schedule_1x.py：定义学习策略，例如leaning_rate、epoch等
…/base/default_runtime.py：定义一些日志等其他信息

3、修改数据加载文件

修改：./mmdetection/configs/base/datasets/voc712.py
使用的是VOC2007数据，因此只要把其中含有VOC2012路径注释即可，修改后的内容如下：

# dataset settings
dataset_type = 'VOCDataset'
data_root = 'data/VOCdevkit/'
img_norm_cfg = dict(
    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
train_pipeline = [
    dict(type='LoadImageFromFile'),
    dict(type='LoadAnnotations', with_bbox=True),
    dict(type='Resize', img_scale=(1000, 600), keep_ratio=True),
    dict(type='RandomFlip', flip_ratio=0.5),
    dict(type='Normalize', **img_norm_cfg),
    dict(type='Pad', size_divisor=32),
    dict(type='DefaultFormatBundle'),
    dict(type='Collect', keys=['img', 'gt_bboxes', 'gt_labels']),
]
test_pipeline = [
    dict(type='LoadImageFromFile'),
    dict(
        type='MultiScaleFlipAug',
        img_scale=(1000, 600),
        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']),
        ])
]
data = dict(
    samples_per_gpu=2,
    workers_per_gpu=2,
    train=dict(
        type='RepeatDataset',
        times=3,
        # dataset=dict(
        #     type=dataset_type,
        #     ann_file=[
        #         data_root + 'VOC2007/ImageSets/Main/trainval.txt',
        #         data_root + 'VOC2012/ImageSets/Main/trainval.txt'
        #     ],
        #     img_prefix=[data_root + 'VOC2007/', data_root + 'VOC2012/'],
        #     pipeline=train_pipeline)),
        # 把含有VOC2012的路径去掉
        dataset=dict(
            type=dataset_type,
            ann_file=[
                data_root + 'VOC2007/ImageSets/Main/trainval.txt',
            ],
            img_prefix=[data_root + 'VOC2007/'],
            pipeline=train_pipeline)),
    val=dict(
        type=dataset_type,
        ann_file=data_root + 'VOC2007/ImageSets/Main/test.txt',
        img_prefix=data_root + 'VOC2007/',
        pipeline=test_pipeline),
    test=dict(
        type=dataset_type,
        ann_file=data_root + 'VOC2007/ImageSets/Main/test.txt',
        img_prefix=data_root + 'VOC2007/',
        pipeline=test_pipeline))
evaluation = dict(interval=1, metric='mAP')

4、修改类别

修改：./mmdetection/configs/base/models/faster_rcnn_r50_fpn.py（个数）
修改：./mmdetection/mmdet/core/evaluation/class_names.py（名字）
修改：mmdetection/mmdet/datasets/voc.py（名字）