Webinar: 'Efficient generative AI for images and videos'

With a text prompt and a click of a button you can now generate brand new 2D images, 3D images, and even videos. But at what cost? While this technology has great potential, it faces challenges because the machine learning models have high computation, large memory costs, and low flexibility. As generative AI adoption grows at record-setting speeds, running it efficiently on device is key, as it brings with it advantages in cost, performance, energy, personalization, and privacy.

In this webinar, you will find out:

Why generative vision has a great potential in image & video editing/creation across enterprise, entertainment, XR, and automotive.​
Why efficient generative vision is important.
What can be achieved with state-of-the-art techniques from Qualcomm AI Research, such as Clockwork, Token Merging, and N-to-3.

Qualcomm AI Research is an initiative of Qualcomm Technologies Inc.

About the speaker:
Amirhossein Habibian
Director of Engineering, Qualcomm AI Research
Amirhossein Habibian currently serves as Director of Engineering at Qualcomm AI Research in Amsterdam, focusing on the development of efficient generative models for images and video. He earned his Ph.D. from the University of Amsterdam in 2016, working on learning multimodal video representations. Amir received the Best Paper Award at ACM Multimedia 2014, the Best Paper Award at ASCI 2015, and winning the TRECVID video search challenge in 2014 and 2015. Since joining Qualcomm in 2015, he has contributed to the development of efficient neural architectures across various domains, including video compression, object detection, semantic segmentation, denoising, and super-resolution. His research findings have been disseminated in computer vision conferences such as CVPR, ICCV and ECCV.

Webinar moderated by Armina Stepan (Qualcomm Technologies Netherlands).