DeepSignals: Predicting Intent of Drivers Through Visual Signals
This addresses a critical safety need for self-driving cars by providing early warning of driver actions, though it is incremental as it applies existing deep learning methods to a specific domain.
The paper tackles the problem of detecting driver intentions like lane changes and stops by recognizing turn signals and emergency flashers in video sequences, achieving high per-frame accuracy in challenging scenarios as demonstrated on over a million frames.
Detecting the intention of drivers is an essential task in self-driving, necessary to anticipate sudden events like lane changes and stops. Turn signals and emergency flashers communicate such intentions, providing seconds of potentially critical reaction time. In this paper, we propose to detect these signals in video sequences by using a deep neural network that reasons about both spatial and temporal information. Our experiments on more than a million frames show high per-frame accuracy in very challenging scenarios.