poloclub/cnn-explainer
Cnn Explainer
cnn-explainer is an interactive web application and educational sandbox designed for visualizing the internal operations and layers of convolutional neural networks. It functions as a tool for understanding how these networks process image data through real-time graphics and interactive visualizations.
The project includes a browser-based environment for training small convolutional neural networks on specific image classes. It also provides a model converter that transforms trained neural network files from backend framework formats into web-compatible versions for browser loading.
The application utilizes scalable vector graphics to render network architecture and feature maps. It implements client-side model inference, training loops, and tensor-based data flow to simulate convolutional operations directly in the browser.
Features
- Neural Network Visualizations - Provides an interactive web application for exploring the internal layers and weights of trained convolutional neural networks.
- Client-Side Training - Implements backpropagation and gradient descent entirely within the browser to train small models.
- Browser-Based Inference - Implements real-time neural network inference directly in the browser using a JavaScript tensor library.
- Vision Model Training - Implements a training process for convolutional neural networks focused on image classification.
- Neural Network Visualization Tools - Provides an interactive web application for visualizing the internal operations and layers of convolutional neural networks.
- Interactive Sandboxes - Offers an interactive sandbox for experimenting with how CNNs process image data through real-time graphics.
- Browser-Based Trainers - Ships a browser-based environment to train small convolutional neural networks on specific image classes.
- Convolutional Neural Network Tutorials - Provides an interactive environment for learning how convolutional neural networks process images and perform operations.
- Interactive Visualizations - Displays internal CNN operations through interactive graphics to help users understand the learning process.
- Neural Architecture Visualizations - Renders network architecture and feature maps using scalable vector graphics for interactive inspection.
- Educational Model Training - Enables training of small convolutional neural networks on specific image classes for educational visualization.
- Model Format Converters - Transforms trained neural network files from backend framework formats into web-compatible versions.
- Model Conversion - Transforms trained backend model files into web-compatible formats for browser-based visualization.
- Cross-Framework Translation - Translates model weights from backend deep learning frameworks into a standardized JSON format for browser execution.
- Computer Vision Tools - An interactive visualization tool for understanding convolutional neural networks.
- Machine-Learning-Frameworks - Interaktive Visualisierung zum Erlernen von Convolutional Neural Networks.
- Perception and Machine Learning - Interactive visualization for convolutional neural networks.
