Introduction

The invention refers to a method for integrating information obtained through active depth sensors, such as, for example, Lidar, or from non- learning-based algorithms, used as a guide, into depth estimation systems based on automatic learning from images  such as stereo matching, etc. The method can find the application in Computer Vision, allowing you to estimate depth in learning-based systems

Technical features

The proposed technology acts on the abstract representation of the observed scene inside the deep learning algorithm as well as a conventional method. In particular, measurements obtained from active sensors (or non-learned stereo algorithms) are used to enhance features inside the deep learning algorithm in correspondence of values consistent with depth measured by the sensor or algorithm. Such formulation is flexible and adaptable to other fusion configurations, such as active sensor together with non-learned stereo algorithm.

Possible Applications

• Integration into autonomous driving systems;
• Industrial applications such as Augmented Reality, Robotics;
• Integration into non-learning-based systems.

Advantages

• Allows for errors correction in the estimation of depth in the presence of adverse conditions;
• Makes the depth estimation process more reliable;
• Greatly improves the vehicle’s navigation capacity and robustness in adverse conditions.