This repo contains an App Designer app for training an image classification deep neural networks in MATLAB. Using this app, you can:
- 🖼️ Import, visualize, and augment data
- ⚡ Quickly transfer learn with the SqueezeNet pretrained network
- 🛠️ Modify pretrained networks for transfer learning with Deep Network Designer
- 📥 Import networks from the workspace
- 🔍 Explain predictions with explainability techniques like Grad-CAM and LIME
- 🧾 Generate MATLAB code for training an image classifier
You can also edit and customize the app code for your own task.
- MATLAB ® R2025a or later
- Deep Learning Toolbox™
- (Optional) Parallel Computing Toolbox™ to train models on GPU
- (Optional) Image Processing Toolbox™ to use the
imageLIMEinterpretability function
Import data and apply standard augmentations.
Choose a pretrained model, edit or build a model in Deep Network Designer, or import a model from the workspace.
Train the model, predict on data, and try out interpretability techniques.
To open the app, run the opening function at the MATLAB command line.
ImageClassifer;
To modify the app, run appdesigner("ImageClassifier.mlapp").
Copyright 2026 The MathWorks, Inc.