OpenMMLab AI实战营 第三课笔记
花朵五分类数据集：https://www.kaggle.com/datasets/alxmamaev/flowers-recognition
进入 mmclassification 目录
In [1]:
import os
os.chdir('mmclassification')
导入工具包
In [2]:
import torch
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
print('device', device)
device cuda:0
下载数据集
In [3]:
!wget https://zihao-openmmlab.obs.cn-east-3.myhuaweicloud.com/20220716-mmclassification/dataset/flower.zip -O data/flower.zip
--2022-07-16 22:34:18-- https://zihao-openmmlab.obs.cn-east-3.myhuaweicloud.com/20220716-mmclassification/dataset/flower.zip
Connecting to 172.16.0.13:5848... connected.
Proxy request sent, awaiting response... 200 OK
Length: 230662310 (220M) [application/zip]
Saving to: ‘data/flower.zip’
data/flower.zip 100%[===================>] 219.98M 27.9MB/s in 7.8s
2022-07-16 22:34:28 (28.3 MB/s) - ‘data/flower.zip’ saved [230662310/230662310]
In [4]:
# 解压
!unzip data/flower.zip -d data >> /dev/null
In [13]:
from PIL import Image
Image.open('data/flower/test/daisy/11023214096_b5b39fab08.jpg')
数据集目录结构
In [21]:
'''
flower
├── classes.txt
├── train.txt
├── val.txt
├── test.txt
├── train
│ ├── daisy
│ ├── dandelion
│ ├── rose
│ ├── sunflower
│ └── tulip
├── test
│ ├── daisy
│ ├── dandelion
│ ├── rose
│ ├── sunflower
│ └── tulip
└── val
├── daisy
├── dandelion
├── rose
├── sunflower
└── tulip
'''
下载 config 配置文件
In [30]:
'''
Model config, which specify the basic structure of the model, e.g. number of the input channels.
Dataset config, which contains details about the dataset, e.g. type of the dataset.
Schedule config, which specify the training schedules, e.g. learning rate.
Runtime config, which contains the rest of details, e.g. log config.
'''
In [11]:
!wget https://zihao-openmmlab.obs.cn-east-3.myhuaweicloud.com/20220716-mmclassification/configs/mobilenet_v2_1x_flower.py -O configs/mobilenet_v2/mobilenet_v2_1x_flower.py
--2022-07-16 22:51:45-- https://zihao-openmmlab.obs.cn-east-3.myhuaweicloud.com/20220716-mmclassification/configs/mobilenet_v2_1x_flower.py
Connecting to 172.16.0.13:5848... connected.
Proxy request sent, awaiting response... 200 OK
Length: 1975 (1.9K) [binary/octet-stream]
Saving to: ‘configs/mobilenet_v2/mobilenet_v2_1x_flower.py’
configs/mobilenet_v 100%[===================>] 1.93K --.-KB/s in 0s
2022-07-16 22:51:45 (8.72 MB/s) - ‘configs/mobilenet_v2/mobilenet_v2_1x_flower.py’ saved [1975/1975]
命令行-训练
In [12]:
!python tools/train.py \
configs/mobilenet_v2/mobilenet_v2_1x_flower.py \
--work-dir work_dirs/mobilenet_v2_1x_flower
/home/featurize/work/MMClassification教程/mmclassification/mmcls/utils/setup_env.py:33: UserWarning: Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
f'Setting OMP_NUM_THREADS environment variable for each process '
/home/featurize/work/MMClassification教程/mmclassification/mmcls/utils/setup_env.py:43: UserWarning: Setting MKL_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
f'Setting MKL_NUM_THREADS environment variable for each process '
2022-07-16 22:51:55,465 - mmcls - INFO - Environment info:
------------------------------------------------------------
sys.platform: linux
Python: 3.7.10 (default, Jun 4 2021, 14:48:32) [GCC 7.5.0]
CUDA available: True
GPU 0: NVIDIA RTX A4000
CUDA_HOME: /usr/local/cuda
NVCC: Cuda compilation tools, release 11.2, V11.2.152
GCC: gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0
PyTorch: 1.10.0+cu113
PyTorch compiling details: PyTorch built with:
- GCC 7.3
- C++ Version: 201402
- Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 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: AVX512
- 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_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
- 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=1.10.0, 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: 0.11.1+cu113
OpenCV: 4.5.4
MMCV: 1.6.0
MMCV Compiler: GCC 9.3
MMCV CUDA Compiler: 11.3
MMClassification: 0.23.1+d2e5054
------------------------------------------------------------
2022-07-16 22:51:55,465 - mmcls - INFO - Distributed training: False
2022-07-16 22:51:55,601 - mmcls - INFO - Config:
model = dict(
type='ImageClassifier',
backbone=dict(type='MobileNetV2', widen_factor=1.0),
neck=dict(type='GlobalAveragePooling'),
head=dict(
type='LinearClsHead',
num_classes=5,
in_channels=1280,
loss=dict(type='CrossEntropyLoss', loss_weight=1.0),
topk=(1, 3)))
dataset_type = 'ImageNet'
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='RandomResizedCrop', size=224, backend='pillow'),
dict(type='RandomFlip', flip_prob=0.5, direction='horizontal'),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='ImageToTensor', keys=['img']),
dict(type='ToTensor', keys=['gt_label']),
dict(type='Collect', keys=['img', 'gt_label'])
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='Resize', size=(256, -1), backend='pillow'),
dict(type='CenterCrop', crop_size=224),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
]
data = dict(
samples_per_gpu=32,
workers_per_gpu=2,
train=dict(
type='ImageNet',
data_prefix='data/flower/train',
pipeline=[
dict(type='LoadImageFromFile'),
dict(type='RandomResizedCrop', size=224, backend='pillow'),
dict(type='RandomFlip', flip_prob=0.5, direction='horizontal'),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='ImageToTensor', keys=['img']),
dict(type='ToTensor', keys=['gt_label']),
dict(type='Collect', keys=['img', 'gt_label'])
],
classes='data/flower/classes.txt'),
val=dict(
type='ImageNet',
data_prefix='data/flower/val',
ann_file='data/flower/val.txt',
pipeline=[
dict(type='LoadImageFromFile'),
dict(type='Resize', size=(256, -1), backend='pillow'),
dict(type='CenterCrop', crop_size=224),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
],
classes='data/flower/classes.txt'),
test=dict(
type='ImageNet',
data_prefix='data/flower/test',
ann_file='data/flower/test.txt',
pipeline=[
dict(type='LoadImageFromFile'),
dict(type='Resize', size=(256, -1), backend='pillow'),
dict(type='CenterCrop', crop_size=224),
dict(
type='Normalize',
mean=[123.675, 116.28, 103.53],
std=[58.395, 57.12, 57.375],
to_rgb=True),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
],
classes='data/flower/classes.txt'))
evaluation = dict(
interval=1,
metric=['accuracy', 'precision', 'f1_score'],
metric_options=dict(topk=(1, )))
optimizer = dict(type='SGD', lr=0.01, momentum=0.9, weight_decay=0.0001)
optimizer_config = dict(grad_clip=None)
lr_config = dict(policy='step', gamma=0.98, step=[1])
runner = dict(type='EpochBasedRunner', max_epochs=2)
checkpoint_config = dict(interval=1)
log_config = dict(interval=100, hooks=[dict(type='TextLoggerHook')])
dist_params = dict(backend='nccl')
log_level = 'INFO'
load_from = 'https://download.openmmlab.com/mmclassification/v0/mobilenet_v2/mobilenet_v2_batch256_imagenet_20200708-3b2dc3af.pth'
resume_from = None
workflow = [('train', 1)]
work_dir = 'work_dirs/mobilenet_v2_1x_flower'
gpu_ids = [0]
2022-07-16 22:51:55,601 - mmcls - INFO - Set random seed to 943425345, deterministic: False
2022-07-16 22:51:55,802 - mmcls - INFO - initialize MobileNetV2 with init_cfg [{'type': 'Kaiming', 'layer': ['Conv2d']}, {'type': 'Constant', 'val': 1, 'layer': ['_BatchNorm', 'GroupNorm']}]
2022-07-16 22:51:55,832 - mmcls - INFO - initialize LinearClsHead with init_cfg {'type': 'Normal', 'layer': 'Linear', 'std': 0.01}
2022-07-16 22:52:02,074 - mmcls - INFO - load checkpoint from http path: https://download.openmmlab.com/mmclassification/v0/mobilenet_v2/mobilenet_v2_batch256_imagenet_20200708-3b2dc3af.pth
2022-07-16 22:52:02,104 - mmcls - WARNING - The model and loaded state dict do not match exactly
size mismatch for head.fc.weight: copying a param with shape torch.Size([1000, 1280]) from checkpoint, the shape in current model is torch.Size([5, 1280]).
size mismatch for head.fc.bias: copying a param with shape torch.Size([1000]) from checkpoint, the shape in current model is torch.Size([5]).
2022-07-16 22:52:02,105 - mmcls - INFO - Start running, host: featurize@featurize, work_dir: /home/featurize/work/MMClassification教程/mmclassification/work_dirs/mobilenet_v2_1x_flower
2022-07-16 22:52:02,105 - mmcls - INFO - Hooks will be executed in the following order:
before_run:
(VERY_HIGH ) StepLrUpdaterHook
(NORMAL ) CheckpointHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
--------------------
before_train_epoch:
(VERY_HIGH ) StepLrUpdaterHook
(LOW ) IterTimerHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
--------------------
before_train_iter:
(VERY_HIGH ) StepLrUpdaterHook
(LOW ) IterTimerHook
(LOW ) EvalHook
--------------------
after_train_iter:
(ABOVE_NORMAL) OptimizerHook
(NORMAL ) CheckpointHook
(LOW ) IterTimerHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
--------------------
after_train_epoch:
(NORMAL ) CheckpointHook
(LOW ) EvalHook
(VERY_LOW ) TextLoggerHook
--------------------
before_val_epoch:
(LOW ) IterTimerHook
(VERY_LOW ) TextLoggerHook
--------------------
before_val_iter:
(LOW ) IterTimerHook
--------------------
after_val_iter:
(LOW ) IterTimerHook
--------------------
after_val_epoch:
(VERY_LOW ) TextLoggerHook
--------------------
after_run:
(VERY_LOW ) TextLoggerHook
--------------------
2022-07-16 22:52:02,105 - mmcls - INFO - workflow: [('train', 1)], max: 2 epochs
2022-07-16 22:52:02,105 - mmcls - INFO - Checkpoints will be saved to /home/featurize/work/MMClassification教程/mmclassification/work_dirs/mobilenet_v2_1x_flower by HardDiskBackend.
2022-07-16 22:52:11,810 - mmcls - INFO - Saving checkpoint at 1 epochs
[>>>>>>>>>>>>>>>>>>>>>>>>>>>>>] 715/715, 354.3 task/s, elapsed: 2s, ETA: 0s2022-07-16 22:52:13,944 - mmcls - INFO - Epoch(val) [1][23] accuracy_top-1: 66.1538, precision: 73.5692, f1_score: 65.5141
2022-07-16 22:52:23,245 - mmcls - INFO - Saving checkpoint at 2 epochs
[>>>>>>>>>>>>>>>>>>>>>>>>>>>>>] 715/715, 360.9 task/s, elapsed: 2s, ETA: 0s2022-07-16 22:52:25,354 - mmcls - INFO - Epoch(val) [2][23] accuracy_top-1: 88.6713, precision: 89.7683, f1_score: 88.7995
用训练得到的图像分类模型，对新图像预测
In [16]:
import matplotlib.pyplot as plt
import mmcv
from mmcls.apis import inference_model, init_model, show_result_pyplot
img = mmcv.imread('data/flower/test/daisy/11023214096_b5b39fab08.jpg')
# img = mmcv.imread('data/cat2.jpg')
# 图像分类模型 config 配置文件
config_file = 'configs/mobilenet_v2/mobilenet_v2_1x_flower.py'
# 图像分类模型 checkpoint 权重文件
checkpoint_file = 'work_dirs/mobilenet_v2_1x_flower/latest.pth'
# 通过 config 配置文件 和 checkpoint 权重文件 构建模型
model = init_model(config_file, checkpoint_file, device=device)
result = inference_model(model, img)
print('类别', result['pred_class'], '置信度', result['pred_score'])
show_result_pyplot(model, img, result)
load checkpoint from local path: work_dirs/mobilenet_v2_1x_flower/latest.pth
类别 daisy 置信度 0.9996930360794067
将训练得到的模型在测试集上预测，获得所有测试集数据的预测结果
In [17]:
!python tools/test.py \
configs/mobilenet_v2/mobilenet_v2_1x_flower.py \
work_dirs/mobilenet_v2_1x_flower/latest.pth \
--out testset_result.json
/home/featurize/work/MMClassification教程/mmclassification/mmcls/utils/setup_env.py:33: UserWarning: Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
f'Setting OMP_NUM_THREADS environment variable for each process '
/home/featurize/work/MMClassification教程/mmclassification/mmcls/utils/setup_env.py:43: UserWarning: Setting MKL_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
f'Setting MKL_NUM_THREADS environment variable for each process '
load checkpoint from local path: work_dirs/mobilenet_v2_1x_flower/latest.pth
[>>>>>>>>>>>>>>>>>>>>>>>>>>>>>] 715/715, 358.6 task/s, elapsed: 2s, ETA: 0s
dumping results to results_flower.json
将训练得到的模型在测试集上预测，获得图像分类评估结果
In [18]:
!python tools/test.py \
configs/mobilenet_v2/mobilenet_v2_1x_flower.py \
work_dirs/mobilenet_v2_1x_flower/latest.pth \
--metrics accuracy precision recall f1_score support \
--metric-options topk=1
/home/featurize/work/MMClassification教程/mmclassification/mmcls/utils/setup_env.py:33: UserWarning: Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
f'Setting OMP_NUM_THREADS environment variable for each process '
/home/featurize/work/MMClassification教程/mmclassification/mmcls/utils/setup_env.py:43: UserWarning: Setting MKL_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
f'Setting MKL_NUM_THREADS environment variable for each process '
load checkpoint from local path: work_dirs/mobilenet_v2_1x_flower/latest.pth
[>>>>>>>>>>>>>>>>>>>>>>>>>>>>>] 715/715, 352.3 task/s, elapsed: 2s, ETA: 0s
accuracy : 88.67
support : 715.0
precision : 89.77
recall : 88.83
f1_score : 88.8