Font-identifier Open-source Font Recognition Model - Free Deployment for Easy and Accurate Font Recognition

Home

Font Identifier

Developed by ariadnak

A font recognition model fine-tuned on ResNet-18, achieving 78.1% accuracy on the test set

Image Classification

Transformers

Open Source License:Apache-2.0 #Font Recognition #Image Classification #ResNet Fine-tuning

Downloads 44

Release Time : 12/4/2023

Model Overview

This model is designed for image classification tasks, specifically for identifying different font types. Fine-tuned based on Microsoft's ResNet-18 architecture, it is suitable for document processing, design assistance, and similar scenarios.

Model Features

High Accuracy

Achieves 78.1% accuracy on the test set, effectively recognizing multiple fonts

Lightweight Architecture

Based on ResNet-18 architecture, maintaining performance while keeping computational overhead low

End-to-End Training

Fine-tuned using a complete training process, including 30 training epochs and linear learning rate scheduling

Model Capabilities

Font Classification

Image Feature Extraction

Document Analysis

Use Cases

Document Processing

Automatic Font Recognition

Identifies font types in scanned documents or images

78.1% accuracy

Design Assistance

Design Material Classification

Automatically classifies and organizes fonts in design material libraries

🚀 font-identifier

This model is a fine - tuned version of microsoft/resnet-18 on the imagefolder dataset, which can be used for image classification tasks.

🚀 Quick Start

This model is a fine - tuned version of microsoft/resnet-18 on the imagefolder dataset. It achieves the following results on the evaluation set:

Loss: 0.8935
Accuracy: 0.7810

📚 Documentation

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e - 05
train_batch_size: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 64
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
7.2836	1.0	344	7.2178	0.0038
6.6696	2.0	689	6.4685	0.0408
5.85	3.0	1034	5.3897	0.1254
5.0457	4.0	1379	4.4771	0.2143
4.3784	5.0	1723	3.6429	0.3242
3.809	6.0	2068	3.1236	0.4031
3.4229	7.0	2413	2.6388	0.4672
2.8977	8.0	2758	2.3279	0.5102
2.78	9.0	3102	2.0974	0.5682
2.4452	10.0	3447	1.8605	0.6027
2.2195	11.0	3792	1.6783	0.6312
2.1097	12.0	4137	1.6049	0.6390
1.9025	13.0	4481	1.4255	0.6912
1.7973	14.0	4826	1.3253	0.7075
1.7647	15.0	5171	1.3030	0.7032
1.6772	16.0	5516	1.1988	0.7210
1.5523	17.0	5860	1.1040	0.7395
1.4821	18.0	6205	1.0786	0.7380
1.3764	19.0	6550	1.0603	0.7471
1.2913	20.0	6895	1.0169	0.7542
1.3479	21.0	7239	0.9999	0.7563
1.3133	22.0	7584	0.9928	0.7594
1.2241	23.0	7929	0.9342	0.7649
1.1651	24.0	8274	0.9283	0.7658
1.1605	25.0	8618	0.9176	0.7720
1.0283	26.0	8963	0.8970	0.7767
1.1211	27.0	9308	0.8983	0.7754
1.1563	28.0	9653	0.8729	0.7801
1.1399	29.0	9997	0.9021	0.7732
1.1715	29.93	10320	0.8935	0.7810

Framework versions

Transformers 4.35.2
Pytorch 2.0.0
Datasets 2.15.0
Tokenizers 0.15.0

📄 License

This model is licensed under the Apache - 2.0 license.

📋 Model Information

Property	Details
Model Type	Fine - tuned version of microsoft/resnet - 18
Training Data	imagefolder
Metrics	accuracy
Results on Test Set	Loss: 0.8935, Accuracy: 0.7810

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