Open Source · 2022 TensorFlow Community Spotlight

ConvNet Runner

A desktop application enabling anyone to train Convolutional Neural Networks without writing a single line of code. Built to democratise machine learning access for researchers, educators, and non-programmers.

View on GitHub ↗ TensorFlow Spotlight ↗
Year
2022
Category
Open Source · ML Tools
Award
TF Community Spotlight
Domain
Computer Vision · CNN

Background

Machine learning has historically required significant programming expertise to access. During my undergraduate studies, I noticed that many researchers and educators who understood the theory of deep learning were blocked by the engineering barrier — they couldn't experiment with CNNs without writing Python code from scratch.

ConvNet Runner was built to close that gap. The idea was simple: expose the full power of TensorFlow's CNN training pipeline through a clean graphical interface, so that anyone — regardless of coding background — could upload a dataset, configure a network architecture, and train a model.

Features

Zero-code CNN training
Train image classification models entirely through the GUI — no Python, no terminal.
Configurable architecture
Adjust layers, filters, pooling, activation functions, and hyperparameters visually.
Custom dataset support
Import any image classification dataset. Automatic preprocessing and train/val splitting.
Live training metrics
Real-time accuracy and loss visualisation during training epochs.

Built with

Python TensorFlow 2.x Keras Tkinter NumPy Matplotlib PIL / Pillow

Impact

TF Community Spotlight
2022 Released & featured
No-code accessibility
T
TensorFlow
@TensorFlow
Community Spotlight: ConvNet Runner by @KhalidAlnujaidi — a desktop app that lets you train CNNs for image classification without any coding. Democratising ML, one GUI at a time.
View original tweet ↗

Related research

This project grew alongside my research on computer vision applications. Work on camel detection and road safety from the same period evolved into two published arXiv papers — Spot-the-Camel (12 citations) and Computer Vision for Camel-Vehicle Collision Mitigation — both building on the CNN knowledge developed during ConvNet Runner.