12-08, 18:30–20:00 (UTC), General Track
This session is designed for those who are curious about Keras and want to learn more about its capabilities for computer vision and stable diffusion. We will start with a refresher on the core deep learning concepts that are essential for understanding Keras. Then, we will dive into a quick introduction to Keras 3 with Jax, using object detection as an example. Next, we will explore how to use Keras CV and Keras 3 together for multi-framework modeling that includes . We will also discuss how to use pre-trained PyTorch models with Keras 3. Finally, we will wrap up with a discussion of stable diffusion, what it is, and how to implement it using Keras 3 and multi-framework modeling.
This session is perfect for anyone who wants to learn more about Keras 3 and its capabilities for computer vision and stable diffusion. Whether you are a beginner or an experienced machine learning practitioner, you will find something valuable in this session.
We will start with a refresher on the core deep learning concepts that are essential for understanding Keras and TensorFlow. This will include topics such as neural networks, convolutional neural networks (CNNs), and backpropagation.
Next, we will dive into a quick introduction to Keras 3 with Jax. Keras 3 is a new library that enables you to have seamless integration with TensorFlow, JAX, and PyTorch workflows, allowing users to combine the best features of each framework.
After that, we will explore how to use Keras CV and Keras 3 together for multi-framework modeling. Keras CV is a library that provides Keras bindings for popular computer vision libraries such as OpenCV and PyTorch. We will show you how to use Keras CV and Keras Core to build a multi-framework model for image classification.
We will also discuss how to use pre-trained PyTorch models with Keras 3. This allows you to take advantage of the large selection of pre-trained models that are available for PyTorch. We will show you how to use a pre-trained PyTorch model to build a Keras Core model for image segmentation.
Finally, we will wrap up with a discussion of stable diffusion, what it is, and how to implement it using Keras Core and multi-framework modeling.
Previous knowledge expected
Ngesa is an Electrical Engineer specializing in Signal Processing and Computer Vision. He started his ML journey as an Intel AI Innovator and later joined Liquid Intelligent Technologies as an IoT Solutions Engineer. He is currently a Device Manager, Safaricom PLC focusing on Cloud & AIoT use cases. He is also an Arm AI Ambassador.
When not building products, he loves to teach and share knowledge with others. He believes ML education should be accessible to all. Recently, he has been working on sensor data analysis from IoT devices to help manage Device lifecycles and drive the right customer experience.
He founded Nairobi AI and leads Machine Learning efforts in communities such as GDG Nairobi and Tiny ML Kenya.