Skip to content

Recapping the AI, Machine Learning and Data Science Meetup — March 21, 2024

We just wrapped up the March ‘24 AI, Machine Learning and Data Science Meetup, and if you missed it or want to revisit it, here’s a recap! In this blog post you’ll find the playback recordings, highlights from the presentations and Q&A, as well as the upcoming Meetup schedule so that you can join us at a future event. 

First, Thanks for Voting for Your Favorite Charity!

In lieu of swag, we gave Meetup attendees the opportunity to help guide a $200 donation to charitable causes. The charity that received the highest number of votes this month was Oceana, which is focused on ocean conservation — protecting and restoring marine life and the world’s abundant and biodiverse oceans. We are sending this event’s charitable donation of $200 to Oceana on behalf of the Meetup members!

Missed the Meetup? No problem. Here are playbacks and talk abstracts from the event.


Visual Anagrams: Generating Multi-View Optical Illusions with Diffusion Models



Images that change their appearance under a transformation, such as a rotation or a flip, have long fascinated students of perception, from Salvador Dalli to M. C. Escher. The appeal of these multi-view optical illusions lies partly in the challenge of arranging visual elements such that they may be understood in multiple different ways. Creating these illusions requires accurately modeling—and then subverting—visual perception. We propose a simple, zero-shot method for obtaining these illusions from off-the-shelf text-to-image diffusion models. During the reverse diffusion process, we estimate the noise from different views of a noisy image. We then combine these noise estimates together and denoise the image. A theoretical analysis suggests that this method works precisely for views that can be written as orthogonal transformations. An important special case of these transformations are permutations of an image’s pixels, which we use to create a variety of “anagram” illusions, such as jigsaw puzzles that can be solved in two different ways.

Speaker: Andrew Owens is an assistant professor at the University of Michigan in the department of Electrical Engineering and Computer Science. Prior to that, he was a postdoctoral scholar at UC Berkeley. He received a Ph.D. in Electrical Engineering and Computer Science from MIT in 2016. He is a recipient of a Computer Vision and Pattern Recognition (CVPR) Best Paper Honorable Mention Award, and a Microsoft Research Ph.D. Fellowship.

Q&A

  • How much data was required to train this kind of diffusion model?
  • Do you see your work as a way to edit images via prompts?
  • How do you create a deception illusion picture using just prompts instead of taking a perfection illusion image as a reference?
  • Are your models trained from scratch or fine tuned from pre-trained ones?
  • How many parameters do these kinds of models have?
  • Do you need to explicitly indicate the direction in which the image is rotated or flipped in the text prompt?
  • Is there a certain number of puzzle pieces that works best for this type of illusion?
  • Is there an upper limit to the number of views you can properly create?
  • Can you also create 3D illusions?

Resource links

Omnidirectional Computer Vision



Omnidirectional cameras are a different camera modality than what we are typically used to, where we sacrifice nice geometry for significantly larger fields-of-view. Ciarán will discuss Omnidirectional cameras, how to represent them geometrically, optical effects of using omnidirectional cameras, neural networks on omnidirectional cameras, and applications that use omnidirectional cameras.

Speaker: Ciarán Eising is an Associate Professor of Artificial Intelligence and Computer Vision at the University of Limerick, Ireland and co-founder of the D2iCE Research Group. Prior to joining the University of Limerick, Ciarán was a Senior Expert of Computer Vision (director level) at Valeo, designing omnidirectional vision algorithms for low-speed vehicle automation.

Q&A

  • What is the input size of the images for the model?
  • Did you differentiate between 360 degrees in one plane and 360 degrees in 3D dimensions?
  • Can you explain in more detail the use of transformers in your research?

Resource links

Illuminating the Underground World with Multimodal Algorithms



Dive deep into the future of underground exploration! This talk introduces a groundbreaking approach that leverages the power of multimodal algorithms, combining advanced computer vision techniques, SLAM algorithms, sensor metadata, and GIS data. By integrating diverse data streams, we unlock unprecedented levels of detail and accuracy in underground inspections, paving the way for safer, more efficient, and insightful subterranean analyses.

Speaker: Adonaí Vera is a Machine Learning Engineer with expertise in computer vision and AI algorithms, specializing in AI solutions for underground inspections using TensorFlow and OpenCV. Recognized by Google as a top TensorFlow developer in Colombia, he is also the founder of a company focused on AI innovations and currently contributes his expertise to Subterra AI.

Q&A

  • How do you determine when the model/algorithm is good enough for use since accuracy is important in these kinds of situations?

Resource links

The Role of AI in Fixing Hiring



From writing job descriptions, to sourcing candidates, to interviews and evaluation – our current hiring practices are often rife with human bias. Even when such bias is unconscious, it can result in expensive mis-hires. This talk explores the types of biases and the pivotal role of AI in mitigating them. We will discuss the common sources of bias in hiring, the current AI landscape that attempts to address these issues and further opportunities.

Speaker: Saurav Pandit is a seasoned AI leader with expertise in natural language understanding, search and language models. He is on the advisory board of AI 2030, an initiative that promotes AI for good.

Resource links

Join the AI, Machine Learning and Data Science Meetup!

The combined membership of the Computer Vision and AI, Machine Learning and Data Science Meetups has grown to over 20,000 members! The goal of the Meetups is to bring together communities of data scientists, machine learning engineers, and open source enthusiasts who want to share and expand their knowledge of AI and complementary technologies.

Join one of the 12 Meetup locations closest to your timezone.

What’s Next?

Up next on April 18 at 10 AM Pacific we have two great speakers lined up!

  • Lumiere: A Space-Time Diffusion Model for Video GenerationHila Chefer at Google and Tel Aviv University
  • Towards Resource Efficient Robust Text-to-Image Generative ModelsMaitreya Patel at Arizona State University

Register for the Zoom here. You can find a complete schedule of upcoming Meetups on the Voxel51 Events page.

Get Involved!

There are a lot of ways to get involved in the Computer Vision Meetups. Reach out if you identify with any of these:

  • You’d like to speak at an upcoming Meetup
  • You have a physical meeting space in one of the Meetup locations and would like to make it available for a Meetup
  • You’d like to co-organize a Meetup
  • You’d like to co-sponsor a Meetup

Reach out to Meetup co-organizer Jimmy Guerrero on Meetup.com or ping me over LinkedIn to discuss how to get you plugged in.

These Meetups are sponsored by Voxel51, the company behind the open source FiftyOne computer vision toolset. FiftyOne enables data science teams to improve the performance of their computer vision models by helping them curate high quality datasets, evaluate models, find mistakes, visualize embeddings, and get to production faster. It’s easy to get started, in just a few minutes.