WiDS Sweden & Modulai Hybrid Meetup 2023

Looking forward to hanging out with you over some vegetarian bowls. The reception at AI Sweden building closes at 17:00, but we will try to make sure someone is available to help you up still (or email women@wids.se if you're waiting). 

Deep learning has been successfully used for various computer vision problems; however, often, models need large amounts of data to solve a specific task effectively. Unfortunately, there is often a bottleneck to accessing labeled data. With the advancement in graphic tools, simulator engines, and generative models, synthetic data generation has become a suitable alternative to approach the lack of data in specific domains.We will talk about the current development in synthetic data generation and how we have used it to build good models for computer vision tasks based primarily on synthetic data. We will go through the different use cases, explain the tools used and give an insight into how we approach a computer vision problem when we lack a large amount of labeled data.

Computer Vision, today, is being widely used for several applications. However, pre-processing and training on large amounts of data can prove to be quite time-consuming and computationally heavy. JAX, an ML framework created by Google, enables a faster run time owing to its capacity to accelerate operations on GPUs and TPUs. The talk will be focused on JAX and give an overview of its application in the realm of Computer VisionDuring the talk, we will talk about the frameworks frequently used to handle large data, such as PyTorch and Tensorflow. We will discuss JAX, its distinctiveness compared to other packages, and its many advantages. Following this, we will delve into some technical details and how JAX fares in comparison to the more commonly used packages. The presentation will end with a hands-on coding demo, where we will create a simple CNN model using the MNIST dataset.

This talk explores the impact of deep learning in medical imaging. Centered on breast cancer - the most common type of cancer among women - we aim to underscore the significance of integrating advanced solutions in healthcare. By leveraging deep learning for breast cancer detection and risk estimation, hospitals can enhance their resource allocation and patient care. We will also discuss the challenges inherent in applying deep learning and share insights from our case study on navigating these issues.

Interest in Generative AI and its usage accelerated this year, with large language models taking most of the spotlight, but computer vision models and companies like Midjourney, Stable Diffusion and others also grabbing big headlines – and everyone’s imagination. In this short panel, we will discuss Generative AI in the field of Computer Vision more specifically. Bring your questions and stories!