Mnist Digit Classification 2022 09 04
M
Mnist Digit Classification 2022 09 04
Developed by farleyknight
This is a MNIST handwritten digit classification model based on the Vision Transformer (ViT) architecture, achieving 99.23% accuracy after fine-tuning on the MNIST dataset.
Downloads 740
Release Time : 9/4/2022
Model Overview
The model uses google/vit-base-patch16-224-in21k as the base model and is fine-tuned on the MNIST handwritten digit dataset, specifically designed for image classification tasks of digits 0-9.
Model Features
High accuracy
Achieves 99.23% classification accuracy on the MNIST test set
ViT-based architecture
Uses the Vision Transformer architecture, which may have better feature extraction capabilities compared to traditional CNNs
Lightweight fine-tuning
Fine-tuned on a pre-trained model with high training efficiency
Model Capabilities
Handwritten digit recognition
Image classification
Digit 0-9 classification
Use Cases
Education
Automated grading of handwritten digit assignments
Used in educational settings to automatically recognize students' handwritten digit answers
Recognition accuracy of 99.23%
Finance
Check digit recognition
Recognize handwritten amount digits on checks
Featured Recommended AI Models
Š 2025AIbase