This repository contains the files related to the LearnOpenCV blog post: Building MobileViT Image Classification Model from Scratch In Keras 3

In the rapidly evolving field of deep learning, the challenge often lies not just in designing powerful models but also in making them accessible and efficient for practical use, especially on devices with limited computational power. This blog post addresses this challenge by focusing on the MobileViT model, a compact yet robust alternative to the larger and more complex Vision Transformers (ViT). Our primary objective is to provide a comprehensive guide to implementing the MobileViT v1 model from scratch using Keras 3, an approach that ensures compatibility across major frameworks like TensorFlow, PyTorch, and Jax.

You can run it on your local system or on Colab.

Want to become an expert in AI? AI Courses by OpenCV is a great place to start.