AIbase
EN
Home
Artwork Scorer
A

Artwork Scorer

Developed by Muinez
This model is a fine-tuned version based on Facebook's ConvNeXtV2 architecture, specifically trained for multi-label classification tasks on Pixiv ranking images
Image Classification
Transformers
Open Source License:Apache-2.0#Pixiv Image Analysis#Multi-label Classification#Like Rate Prediction
Downloads 32
Release Time : 12/29/2023

Model Overview

A vision-based multi-label classification model fine-tuned on 120,000 Pixiv ranking images, using the standardized ratio of likes to views as the training target

Model Features

Pixiv Data Fine-tuning
Optimized specifically for anime-style images, trained with real user interaction data from Pixiv rankings
Multi-label Classification
Capable of simultaneously identifying multiple tags/features in an image
Standardized Rating Prediction
Training target is the standardized ratio of likes to views, not simple classification

Model Capabilities

Anime Image Analysis
Multi-label Classification
User Preference Prediction

Use Cases

Content Recommendation
Anime Work Popularity Prediction
Predict potential user interaction popularity of works based on image content
Content Moderation
Automatic Tag Generation
Automatically generate descriptive tags for anime images
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase