Transformers models are already considered the highest standard in NLP (Photo: Dreamstime)
Computer vision is one of the hottest fields today, and it has various uses – from medicine to autonomous vehicles. The new and undisputed stars in the field are the Transformers models, which have become a competitive alternative to deep convolutional networks (CNNs) – which are widely used in various image recognition tasks and their reputation is in place. But Transformers models are the hot trend of 2022, and today they are already considered the standard in natural language processing (NLP) tasks.
But wait, what exactly is Transformer and how does it work? In machine learning, Transformer is a deep learning model that uses Attention mechanisms, weighting each part of the input data. The Transformers consist of multiple Self-Attention layers. These models are mainly used in the artificial intelligence subfields of natural language processing (NLP) and computer vision (CV). The Transformers hold strong promises for a generic learning method that can be applied to different data modalities, including recent breakthroughs in computer vision in achieving maximum accuracy with better parameter efficiency.
The use of the autonomous vehicle
A new General Motors meetup will focus on the world of autonomous vehicles and deal with the use of Vision transformers for detection tasks. The event, “Vision Transformers for 3D Perception”, will take place on November 21 at the Dizzy Prishdon Bar in Tel Aviv (7:00 PM). It is intended for students of exact sciences and engineering, researchers and developers in the fields of machine learning and computer vision, and of course – for workers in the field of autonomous vehicles.
The event will present the most up-to-date articles in the field and the leading methods in it, and will focus on the use of Transformers for the purpose of unifying and projecting several cameras for map display (Bird-Eye-View) and 3D object recognition in the autonomous vehicle. The speakers are Natalie Efrat Sela, Senior algorithm developer at General Motors, and Oded Fischler, Perception Algorithms Team Lead at General Motors.
This is a topic that concerns many companies dealing with the challenge of 3D object detection for autonomous driving, companies dealing with computer vision in general and those dealing with a 3D reconstruction challenge. The Transformers have replaced the common architectures in most systems being studied today, so an understanding of this component is interesting and relevant to a wide range of applications in the field of AI.
For more details and registration
About General Motors
General Motors’ development center in Israel was established in 2008 and is a significant part of the company’s growth and the automotive revolution it is leading. The company employs more than 650 engineers who are engaged in fulfilling the company’s vision of a world without accidents, without emissions and without loads, by developing core technologies for autonomous, electric and connected vehicles – technologies that will affect the lives of all of us and change the way we move from place to place.
General Motors Israel is very diverse in its areas of activity and faces challenges, tasks and ambitious goals that will affect the future automotive world mainly (but not only) around machine learning for advanced levels of autonomy, electric and connected vehicles and user experience.
About WizeTime
A series of digital events about technology, entrepreneurship and everything in between. As part of the series, we host leading figures in the high-tech industry and the local entrepreneurial scene. The activity is in collaboration with Gigtime, the leading media body in Israel in the fields of technology and entrepreneurship; and the social organization Wize, one of the leading social production bodies in Israel, which annually produces about 400 events, webinars and workshops with the aim of making science and knowledge accessible to the general public. WizeTime’s activities are carried out in cooperation and with the support of the leading high-tech companies and investment bodies in Israel, and receives wide exposure through the communities of its partners: Gigtime (more than 700 thousand unique surfers per month) and Wize (more than 150 thousand uniques through the organization’s media properties).
Looking to produce or participate in similar events? Contact us by email: [email protected]