/CelebA_classification

Primary LanguagePython

CelebA_classification

Description

This is classification gender.

Requirements

  • Torch version:1.12.0+cu113
  • cuda version: 11.3
  • cudnn version:8302

Dataset

https://www.kaggle.com/datasets/jessicali9530/celeba-dataset

Usage

Before the train you should put dataset in train folder.
example dir is :

* dataset
 ├── train
    ├── female
    ├── male
 ├── val
    ├── female
    ├── male

Then you can train model in distributed settings

python train.py --path PATH

  • args
--lr              : default=0.001
--momentum        : default=0.9
--epochs          : default=25
--path            : dataset path
-- pharse         : default=['train','val']
--batch_size      : default=4
--num_workers     : default=4

After learning, checkpoints are stored in the 'models' file.

Test

python test.py --ckpt CHECKPOINT --path PATH

  • args
--ckpt            : trained model(checkpoint) path
--path            : test image path

test