Auramask-ensemble-moon Open-source Image Conversion Model - Beautiful and Practical 2D Image Processing

Auramask Ensemble Moon

Developed by logasja

This model employs an improved VNet architecture for 2D image processing, focusing on image-to-image translation tasks with adversarial and aesthetic optimization features.

Image Generation Open Source License:Gpl-3.0 #Adversarial Image Enhancement #Aesthetic Quality Optimization #Facial Feature Preservation

Downloads 17

Release Time : 3/2/2025

Model Overview

A Keras-implemented improved VNet architecture designed for high-quality image-to-image translation, with special emphasis on preserving facial features and enhancing aesthetic quality.

Model Features

Multi-feature Loss Optimization

Combines ArcFace and VGG-Face feature extractors to ensure facial feature preservation

Aesthetic Quality Assessment

Optimizes output image aesthetic quality using TopIQ-FR and IQASSIMC metrics

Adversarial Training

Model tags indicate adversarial characteristics, potentially enhancing robustness to input perturbations

Efficient Architecture Design

Utilizes improved VNet architecture supporting 256×256 resolution input with batch size 32

Model Capabilities

Image Style Transfer

Facial Feature Preservation

Aesthetic Quality Enhancement

Adversarial Image Generation

Use Cases

Image Enhancement

Portrait Photo Beautification

Enhances aesthetic quality of portrait photos while preserving facial features

Output images improve aesthetic scores while maintaining identity characteristics

Creative Design

Artistic Filter Application

Applies aesthetically creative filters to images

Generates images with artistic styles while preserving original content

🚀 Keras Image-to-Image Model

This is an image-to-image model built with Keras. It addresses aesthetic and quality aspects in image processing, using adversarial techniques and specific filters. It offers high - quality image transformation based on 2D input/output.

🚀 Quick Start

The training logs of this model can be found here.

✨ Features

Tags: This model is tagged with adversarial, aesthetic, quality, and filter, indicating its application in relevant fields.
Metrics: It uses metrics such as TopIQ - FR, ArcFace Cosine Distance, and VGGFace2 Cosine Distance to evaluate performance.
Datasets: Trained on the logasja/FDF dataset.
Base Models: Built upon vnet, logasja/ArcFace, and logasja/VGGFace.

📦 Installation

The README doesn't provide specific installation steps, so this section is skipped.

💻 Usage Examples

The README doesn't provide code examples, so this section is skipped.

📚 Documentation

Model Description

This model uses a modified vnet for 2D input/output implemented here with the following configuration.

{
    "activation": "ReLU",
    "batch_norm": false,
    "filter_num": [
        128,
        256,
        512,
        1024,
        1024
    ],
    "n_labels": 3,
    "output_activation": "tanh",
    "pool": false,
    "res_num_ini": 1,
    "res_num_max": 3,
    "unpool": false
}

{
    "alpha": 0.0001,
    "batch": 32,
    "epochs": 500,
    "epsilon": 1,
    "input": "(256, 256)",
    "losses": {
        "FEAT_ArcFace": {
            "d": "cosine_similarity",
            "f": "ArcFace",
            "name": "FEAT_ArcFace",
            "reduction": "sum_over_batch_size",
            "threshold": 0.68,
            "weight": 0.05
        },
        "FEAT_VGG-Face": {
            "d": "cosine_similarity",
            "f": "VGG-Face",
            "name": "FEAT_VGG-Face",
            "reduction": "sum_over_batch_size",
            "threshold": 0.68,
            "weight": 0.05
        },
        "IQASSIMC": {
            "lower_better": false,
            "name": "IQASSIMC",
            "reduction": "sum_over_batch_size",
            "weight": 0.5
        },
        "TopIQ": {
            "full_ref": true,
            "lower_better": false,
            "name": "TopIQ",
            "reduction": "sum_over_batch_size",
            "score_range": "~0, ~1",
            "weight": 0.5
        }
    },
    "mixed_precision": true,
    "optimizer": {
        "amsgrad": false,
        "beta_1": 0.9,
        "beta_2": 0.999,
        "clipnorm": null,
        "clipvalue": null,
        "ema_momentum": 0.99,
        "ema_overwrite_frequency": null,
        "epsilon": 1e-07,
        "global_clipnorm": null,
        "gradient_accumulation_steps": null,
        "learning_rate": 9.999999747378752e-05,
        "loss_scale_factor": null,
        "name": "adamw",
        "use_ema": false,
        "weight_decay": 0.004
    },
    "seed": "BIIIIIGSTRETCH",
    "testing": 0.01,
    "training": 0.99
}

Model Architecture Plot

🔧 Technical Details

The README doesn't provide in - depth technical details (less than 50 words of specific technical description), so this section is skipped.

📄 License

This model is licensed under the gpl - 3.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご