Computer Vision News - November 2018

Project Another challenge is due to the need for a very quick output: in order to provide a useful alert, the input collected by the camera must be processed by a local CPU via a relatively simple deep learning system; a more complex neural network would require a GPU, which would significantly increase the cost of the system. The network must be trained to know the status of driver attention at any time and to generate an alert if needed. The algorithmic work involves techniques of pose detection , gaze detection , face detection , hand position and gesture detection among other advanced computer vision techniques. Only a deep learning network can efficiently integrate the input information coming from the detection algorithms. The camera is placed conveniently on the dashboard and directed towards the driver. We are able to detect all possible scenarios with no need for an additional camera inside the car for validation. RSIP Vision has already performed a large number of projects in the automotive fields. Do you have a project around ADAS (Advanced driver-assistance systems) or autonomous driving ? Before you start any work on your system, talk to RSIP Vision’s experts! This is another example of how RSIP Vision’s deep learning algorithms are able to provide a great solution to a real world problem. 13 Computer Vision News by RSIP Vision