August Virtual Computer Vision Meetup

When

August 10, 2023 – 10:00 AM PDT (1:00 PM EDT)

Where

Virtual / Zoom

Agenda

• Neural Congealing: Aligning Images to a Joint Semantic Atlas – Dolev Ofri-Amar at Weizmann Institute of Science
• Advancing Personalized Medicine and Radiotherapy through AI-Enabled Computer Vision – Roushanak Rahmat, PhD, AI Researcher
• A Practical Approach to Deep Learning for Computer Vision with Tensorflow 2 – Folefac Martins at Vinsight and ML Instructor

Neural Congealing: Aligning Images to a Joint Semantic Atlas

Presenting Neural Congealing — a zero-shot self-supervised framework for detecting and jointly aligning semantically-common content across a given set of images. Our approach harnesses the power of pre-trained DINO-ViT features to learn: (i) a joint semantic atlas — a 2D grid that captures the mode of DINO-ViT features in the input set, and (ii) dense mappings from the unified atlas to each of the input images. We derive a new robust self-supervised framework that optimizes the atlas representation and mappings per image set, requiring only a few real-world images as input without any additional input information (e.g., segmentation masks). We design our losses and training paradigm to account only for the shared content under severe variations in appearance, pose, background clutter or other distracting objects, and demonstrate results on a plethora of challenging image sets including sets of mixed domains (e.g., aligning images depicting sculpture and artwork of cats), sets depicting related yet different object categories (e.g., dogs and tigers), or domains for which large-scale training data is scarce (e.g., coffee mugs).

Speaker: Dolev Ofri-Amar is an MSc graduate from the Computer Science and Mathematics department at the Weizmann Institute of Science. Her interests are in computer vision and deep learning, mainly focused on image and video analysis and synthesis.

Advancing Personalized Medicine and Radiotherapy through AI-Enabled Computer Vision

In this talk, we will explore the transformative role of computer vision and AI in the realm of medical imaging, focusing on its applications in personalized medicine and radiotherapy. We will delve into cutting-edge research, open source tools, and real-world use cases that demonstrate the potential of AI to enhance diagnostics, treatment planning, and outcome prediction. The presentation will showcase how computer vision techniques coupled with deep learning algorithms are enabling precise and personalized care for patients, revolutionizing the field of healthcare.

Speaker: Roushanak Rahmat, PhD is an accomplished AI scientist with a PhD in AI from Heriot-Watt University. She specializes in computer vision, medical imaging, and data science, developing advanced algorithms and models for healthcare applications. With expertise in deep learning, she has contributed significantly to AI projects in the healthcare sector, particularly in improving radiotherapy treatment. Roushanak is passionate about using AI to revolutionize healthcare and actively shares her knowledge as a public speaker, Women Techmaker ambassador, and through her Medium blog and YouTube channel.

A Practical Approach to Deep Learning for Computer Vision with Tensorflow 2

A detailed walkthrough of Neuralearn’s Deep Learning for Computer Vision course. We shall discuss at a high level how modern deep learning algorithms can be used in solving computer vision tasks using tools like Tensorflow 2, Hugging Face, Onnx, FastAPI, Weights and Biases and Albumentations, going from the basics of Machine Learning to deploying working computer vision solutions.

In this course, we lay much emphasis on practice, while explaining the theory behind the different algorithms we use. Learners from different backgrounds can easily follow along since efforts are made to explain every concept as clearly and concisely as possible. Because in recent times, deep learning is usually stereotyped as a math-heavy field, we explain in simple terms every math concept, so that learners who aren’t from a math-related background can start building real-world solutions easily.

Given that our focus is mainly on practice, we work on several projects in this course including a car price predictor, a malaria disease classifier, a human emotions detector, an object detector, a digit generator, an image segmenter, a people counter and an image generator.

Speaker: Folefac Martins is an MSc graduate from the Electrical and Telecoms Engineering department at the National Advanced School of Engineering, Polytechnique Yaounde. His interests are in deep learning, helping people realize their potential and entrepreneurship.

Don’t Forget

• Voxel51 will make a donation on behalf of the Meetup members to the charity that gets the most votes this month.
• Can’t make the date and time? No problem! Just make sure to register here so we can send you links to the playbacks.