https://github.com/02loveslollipop/genshinimageclassifier

A convnet trained with Genshin Impact images for character recognition
https://github.com/02loveslollipop/genshinimageclassifier

cnn-classification convnet genshin-impact python pytorch-cnn

Last synced: 9 days ago
JSON representation

A convnet trained with Genshin Impact images for character recognition

• Host: GitHub
• URL: https://github.com/02loveslollipop/genshinimageclassifier
• Owner: 02loveslollipop
• Created: 2024-05-04T00:59:55.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2024-09-12T16:13:56.000Z (5 months ago)
• Last Synced: 2024-11-22T23:24:52.966Z (2 months ago)
• Topics: cnn-classification, convnet, genshin-impact, python, pytorch-cnn
• Language: Jupyter Notebook
• Homepage:
• Size: 12.8 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

        
# Genshin Image Classifier

> A simple image classifier trained on 6 classes of Genshin Impact characters, using a custom Convolutional Neural Network (CNN) architecture.

## Introduction

This project is a simple image classifier trained on 6 classes of Genshin Impact characters. The dataset was obtained from web scraping images from [Pixiv.net](https://www.pixiv.net/) using the tags of each 6 characters. The dataset consists of 9590 training images and 9590 test images, with each class having around 1600 images. The images were preprocessed resized to 128x128 pixels and normalized. The model was trained using a custom Convolutional Neural Network (CNN) architecture. The model was trained for 250 epochs with a batch size of 64. The model achieved an accuracy of 0.75 on the validation set.

## Bias

As the images were obtained by web scraping by the tags of each character, the dataset may contain bias towards the images that despite being tagged with the character's name, may not actually be the character, either because the first image was not the character or the image was not tagged correctly. Also some images may contain multiple characters, which may cause the model to predict the wrong character. 

## Model

TODO: Add model architecture and training details

## Requirements

- A GPU with CUDA support

- Python 3.11

- CUDA 12.1

- Conda

## Installation

1. Clone the repository

```bash

git clone https://github.com/02loveslollipop/GenshinImageClassifier

```	

2. Create conda environment

```bash

conda env create -f environment.yml

```

3. Activate conda environment

```bash

conda activate genshin-image-classifier

```