Deep learning speed camera

After many years of doing both algorithmic research and software development in industry, Dr Violet Snell started a PhD in computer vision at the Centre for Vision, Speech and Signal Processing (CVSSP) in 2010, under the supervision of Josef Kittler and Bill Christmas.

By 2014 she had also completed a lecturing qualification, and taught computer science until the summer of 2015. These days she is developing deep-learning image recognition systems for traffic monitoring at the UK office of Jenoptik Traffic Solutions. She has been involved in mentoring and recruitment of graduate engineers and trainees for nearly two decades, and is an industrial mentor with the Researcher Development Programme at Surrey.

Abstract

Come and find out about real-life applications of cutting-edge deep convolutional networks. Jenoptik is a globally operating integrated photonics group which is present in more than 80 countries. The UK part of their Traffic Solutions division is a world leader in intelligent transportation systems, providing products and solutions based on automatic number plate recognition (ANPR) for applications such as speed and red light enforcement or security and access control. They are currently embarked on a major push to leverage modern machine learning methods in their extensive product lines, to increase the capability and effectiveness of these solutions.

Tracing the full life cycle of a deep learning project, from collecting the right data, through iterative network refinement, to deployment on edge devices, this talk will cover both existing image recognition tasks and future challenges in detection, retrieval and measurement for traffic monitoring applications. The broader Jenoptik group is also exploring the potential of deep convolutional auto-encoders for anomaly detection in industrial inspection systems, with further applications likely in the near future.

Join us to see first-hand how deep convolutional networks are being used today to improve safety and security on our roads; it might be an offence not to…