Drive Now Pay Later Intensive Driving Courses and Driving Lessons

Watch on YouTube here: Drive Now Pay Later Intensive Driving Courses and Driving Lessons
Via https://www.youtube.com/channel/UCZbg1-q0ZTaN5-WWm-rJovA/videos

Pedestrians, bad weather and cyclists created in a virtual world so AI cars can learn to drive

It may look like a video game, but the new computer simulation developed by a team of researchers in Barcelona could one day train autonomous cars to be better drivers.

Called ‘Synthia,’ the program creates a virtual city complete with pedestrians, traffic signs and other components of an urban environment, automatically annotated at the pixel-level.

This allows for a more efficient method of training AI systems, and can be used to teach them to recognize and behave in response to the less predictable aspects of city driving, like a nearby cyclist or adverse weather.

 

A new computer simulation could one day train autonomous cars to be better drivers. Called ‘Synthia,’ the program creates a virtual city complete with pedestrians, traffic signs and other components of an urban environment, automatically annotated at the pixel-level

Described in a recently published paper, Synthia was developed by the Computer Vision Center’s Advanced Driver Assistance Systems (ADAS) research group.

The researchers used more than 213,400 images of both still and video sequences in a virtual city to create the synthetic dataset, Synthetic collection of Imagery and Annotations (Synthia).

In the dataset, the different elements of the city are annotated down to the pixel-level for precise recognition.

Using the Unity development platform to create the urban simulation, they also built a virtual car equipped with multiple cameras.

The data set contains photo-realistic frames generated at up to eight viewpoints per location, along with a depth map.

Synthia was then used to train a deep convolutional neural network (DC-NN) to recognize 11 common classes within a driving scene: sky, building, road, sidewalk, fence, vegetation, pole, car, sign, pedestrian, and cyclist.

They even included the four different seasons and ‘dynamic illumination’ to indicate different times of day, allowing for the simulation of sunny or cloudy days, dusk, and shadows from cloud cover.

While training the networks on synthetic data produced ‘good results’ when recognizing roads, buildings, cars, and pedestrians, accuracy was ‘dramatically boosted’ when combined with real-world data, the researchers wrote.

Original Article:

http://www.dailymail.co.uk/sciencetech/article-3651296/Welcome-Synthia-City-Virtual-world-complete-pedestrians-bad-weather-cyclists-created-AI-cars-learn-drive.html