General
PromptBeginner5 minmarkdown
<h1 align="center">
<a href="https://prompts.chat">
5
General
Qt MNIST Neural Network
English | [Русский](README-RU.md)
promptBeginner5 min to valuemarkdown
0 views
Jan 14, 2026
Sign in to like and favorite skills
Qt MNIST Neural Network
English | Русский
A Qt-based implementation of a neural network for handwritten digit recognition (MNIST-like). To be honest, I'm not entirely sure how it works, but somehow it manages to recognize digits!
Interface
Features
- Draw digits using mouse
- Real-time digit recognition
- Two activation functions:
- Sigmoid
- ReLU
- Pre-trained network weights included
- Training capability
- Error calculation
- Save/Load network weights
Technical Details
The project implements a simple neural network with:
- Input layer (784 neurons - 28x28 pixels)
- Hidden layer (configurable)
- Output layer (10 neurons - digits 0-9)
Pre-trained Models
,sigmoid train.nnw
,sigmoid test.nnw
- Networks trained with Sigmoid activationsigmoid smol.nnw
,ReLU train.nnw
,ReLU test.nnw
- Networks trained with ReLU activationReLU smol.nnw
Building and Running
Requirements
- Qt 5.x or higher
- C++ compiler with C++11 support
- QtCreator (recommended)
Build Steps
- Open
in QtCreatorQt_MNIST.pro - Configure project
- Build and run
Usage
- Launch the application
- Draw a digit in the drawing area
- The network will attempt to recognize it
- Use the progress bars to see confidence levels for each digit
- Clear the drawing area to try another digit
Known Features (or Bugs? Who knows!)
- Sometimes the network is very confident about wrong answers
- Drawing style significantly affects recognition
- The training process is... let's say "mysterious"
License
This project is provided as-is for educational purposes. Use at your own risk - the neural network might have its own opinions about your handwriting!