Sign in | Register to Bookmark

Machine Leaning algorithms performance comparison between thermal and visible imaging

With the merging demand and importance of exceptionally reliable safety critical systems in autonomous driving systems – requirements in safety regulations have challenged current AEB (Automatic Emergency Braking)- systems to their limits. More sensors are added to create a complete sensor suite to cover all possible upcoming situations and to meet the increasing challenging safety regulations. One of the arising technologies which could add an asset to functionality up to replacing some expensive sensors is infrared technology
In this article it will be shown that with similar amount of data, the indicators like accuracy or detection rate are very similar on both thermal and visible imaging under good environment condition, but thermal imaging shows superior performance at low visibility (night, fog) and other conditions. To do a proper comparison an extensive DATASET has been acquired in the scope of HELIAUS project using state of the art Thermal Cameras, Visible range cameras and other reference sensors. The dataset has been labelled for Vulnerable Road Users (VRU) and other vehicles. Finally, deep learning models for each thermal and visible images and detection performance has been evaluated.

Want to watch this session?

You must be logged in and have an AutoSens Brussels 2022 pass to watch this session.

Please either login or create an account.

Already registered? Please Log in

Hear from:

Carles Person

Carles Person
Head of ADAS Project Management
DENSO ADAS Engineering Services GmbH  

Scroll to Top