Classification Pipeline in Computer Vision (Pytorch)
CV_classification
├── datasets
│ ├── __init__.py
│ ├── augmentation.py
│ └── factory.py
├── models
│ ├── __init__.py
│ ├── resnet.py
│ └── loss.py
├── log.py
├── main.py
├── train.py
├── run.sh
├── requirements.txt
├── README.md
└── LICENSE
- Set seed
- Make directory to save results
- Build model
- Build dataset with augmentations
- Train dataset
- Validation dataset (optional)
- Test dataset
- Make dataLoader
- Define optimizer (model parameters)
- Define loss function
- Training model
- Checkpoint model using evaluation on validation dataset
- Log training history using
logging
orwandb
in save folder
- Testing model
run.sh
dataname=$1
num_classes=$2
opt_list='SGD Adam'
lr_list='0.1 0.01 0.001'
aug_list='default weak strong'
bs_list='16 64 256'
for bs in $bs_list
do
for opt in $opt_list
do
for lr in $lr_list
do
for aug in $aug_list
do
# use scheduler
echo "bs: $bs, opt: $opt, lr: $lr, aug: $aug, use_sched: True"
EXP_NAME="bs_$bs-opt_$opt-lr_$lr-aug_$aug-use_sched"
if [ -d "$EXP_NAME" ]
then
echo "$EXP_NAME is exist"
else
python main.py \
--exp-name $EXP_NAME \
--dataname $dataname \
--num-classes $num_classes \
--opt-name $opt \
--aug-name $aug \
--batch-size $bs \
--lr $lr \
--use_scheduler \
--epochs 50
fi
# not use scheduler
echo "bs: $bs, opt: $opt, lr: $lr, aug: $aug, use_sched: False"
EXP_NAME="bs_$bs-opt_$opt-lr_$lr-aug_$aug"
if [ -d "$EXP_NAME" ]
then
echo "$EXP_NAME is exist"
else
python main.py \
--exp-name $EXP_NAME \
--dataname $dataname \
--num-classes $num_classes \
--opt-name $opt \
--aug-name $aug \
--batch-size $bs \
--lr $lr \
--epochs 50
fi
done
done
done
done
example
bash run.sh CIFAR10 10
Argument | Description | Default | Possible value |
---|---|---|---|
use_wandb | Wandb 사용 여부 | True | True,False |
use_cm | Confusion metrix 사용 여부 | True | True,False |
entity | Wandb 엔티티 명 | "connect-cv-04" | --- |
project_name | Wandb 프로젝트 명 | "Image_classification_mask" | --- |
exp_name | 실험명 | "exp" | --- |
exp_num | 실험 번호 | 0 | --- |
user_name | 실험자 | "my_name" | "KDH","KJY","HJH","KDK" |
Argument | Description | Default | Possible value |
---|---|---|---|
datadir | input 경로 | '../input | --- |
train_file | train csv 이름 | "train.csv" | --- |
valid_file | valid csv 이름 | "valid.csv" | --- |
transform | Transform 목록 | ['resize','randomrotation', 'totensor', 'normalize'] | --- |
seed | Random seed | 223 | --- |
model_name | Model_names | “CustomModel” | --- |
model_param | Model_names | {pretrained : True, backbone : "resnet18"} | --- |
num_classes | Class 개수 | 18 | --- |
batch_size | Batch size | 32 | --- |
opt_name | Optimizer | "Adam" | "Adam" |
loss | loss 종류 | "crossentropy" | "crossentropy","focalloss","f1loss","bceloss","mseloss" |
loss_param | loss parm | "미정" | --- |
lr | learning rate | 5e-6 | --- |
lr_sheduler | "Learning rate scheduler | "StepLR" | "StepLR","ReduceLROnPlateau" |
lr_sheduler_param | Lr scheduler parameter | "미정" | --- |
weight_decay | Weight Decay | 5e-4 | --- |
epochs | epoch | 100 | --- |
savedir | 모델 저장 위치 | "./checkpoint" | --- |
grad_accum_steps | --- | 1 | --- |
mixed_precision | --- | "fp16" | --- |
patience | Early Stopping | 100 | --- |
Thanks goes to these wonderful people (emoji key):
김대희 💬 🚧 💻 👀 🤔 |
김동규 💻 🚧 💬 🔬 🔧 |
김준영 💻 🚇 🤔 📆🔬 |
황준하 💻 🤔 💬 🎨 📖 |
This project follows the all-contributors specification. Contributions of any kind welcome!