Anime vs Reality Classification

This project simply classifies whether image is Anime or Reality

Environments

• Python 3.10.9
• torch 1.13.1
• torchvision 0.14.1

Install requirements

pip install -r requirements.txt

Data

• I get my data from various resource on internet. You can use your own data or crawl from internet.

• Data in this format

|-- data
    |-- train
    |   |-- class 1
    |   |-- class 2
    |   `-- ...
    `-- valid
        |-- class 1
        |-- class 2
        `-- ...

Config

Modify config in ./cfg/config.yaml or create your own .yaml config file with the same format.

Train

Simply run

python train.py --cfg ./cfg/config.yaml

Experiment Results

Some experiment results

Model	Accuracy	Confusion Matrix	Pretrained	Model size
SqueezeNet1.1	99.34%		Model	4.74MB
EfficientNet V2 Small	99.71%		Model	82.74MB

You can download weight file above and put in weights folder and run inference

python infer.py

Some inference results

Anime	Reality

You can try on your own 😉