Looking at the right stuff: Guided semantic-gaze for autonomous driving
This addresses the problem of safe and efficient autonomous driving by improving attention prediction, though it appears incremental as it builds on existing saliency methods.
The paper tackles predicting driver attention in autonomous driving by proposing SAGE, which combines human gaze with scene semantics, and SAGE-Net, which further incorporates depth, speed, and pedestrian intent. The result is that SAGE outperforms existing techniques in 87.5% of cases across four saliency algorithms without extra computational overhead.
In recent years, predicting driver's focus of attention has been a very active area of research in the autonomous driving community. Unfortunately, existing state-of-the-art techniques achieve this by relying only on human gaze information, thereby ignoring scene semantics. We propose a novel Semantics Augmented GazE (SAGE) detection approach that captures driving specific contextual information, in addition to the raw gaze. Such a combined attention mechanism serves as a powerful tool to focus on the relevant regions in an image frame in order to make driving both safe and efficient. Using this, we design a complete saliency prediction framework - SAGE-Net, which modifies the initial prediction from SAGE by taking into account vital aspects such as distance to objects (depth), ego vehicle speed, and pedestrian crossing intent. Exhaustive experiments conducted through four popular saliency algorithms show that on $\mathbf{49/56\text{ }(87.5\%)}$ cases - considering both the overall dataset and crucial driving scenarios, SAGE outperforms existing techniques without any additional computational overhead during the training process. The augmented dataset along with the relevant code are available as part of the supplementary material.