Nvidia CES 2016 press conference

Nvidia is kicking off CES 2016 with a press conference expected to focus on automotive and graphics technology.

3rd & 7 37yd

wayne.cunningham 1/5/2016 2:19:40 AM
Nvidia GPUs accelerate the progress of Deep Learning
Lori Grunin 1/5/2016 2:20:14 AM
Lori Grunin 1/5/2016 2:21:56 AM
wayne.cunningham 1/5/2016 2:22:00 AM
Jen-Hsun shows how Deep Learning works into a variety of Nvidia products
Lori Grunin 1/5/2016 2:22:40 AM
wayne.cunningham 1/5/2016 2:23:46 AM
Your car can go out in the world, encounter new situations, then upload the results to the cloud, sharing its experience with every other car.
wayne.cunningham 1/5/2016 2:24:47 AM
Jen-Hsun says the goal is to make Nvidia GPUs essential to every Deep Learning network and application in the world.
wayne.cunningham 1/5/2016 2:25:16 AM
Nvidia developed Drivenet over the past few months
Lori Grunin 1/5/2016 2:25:25 AM
wayne.cunningham 1/5/2016 2:25:38 AM
It's Nvdiia's own neural network designed specifically for cars.
wayne.cunningham 1/5/2016 2:26:00 AM
Nvidia is training this network to identify things in the world.
wayne.cunningham 1/5/2016 2:26:39 AM
According to the graph shown, Nvidia has improved Drivenet hugely over five months
Lori Grunin 1/5/2016 2:27:06 AM
wayne.cunningham 1/5/2016 2:27:13 AM
The graph represents how accurately Drivenet can identify objects from visual imagery
wayne.cunningham 1/5/2016 2:27:31 AM
Mike Huston, senior engineer for mobile and cloud computing at Nvidia, is taking the stage to demo Drivenet
Lori Grunin 1/5/2016 2:28:06 AM
wayne.cunningham 1/5/2016 2:28:19 AM
We're seeing a car drive down a road, recognizing other cars parked and on the road
Lori Grunin 1/5/2016 2:28:23 AM
Mike Houston, principal engineer for mobile and cloud computing at Nvidia
Lori Grunin 1/5/2016 2:28:38 AM
wayne.cunningham 1/5/2016 2:29:37 AM
Jen-Hsun is clarifying that the training to allow this capability took 120 million images
wayne.cunningham 1/5/2016 2:30:10 AM
Sorry, 120 million images, and it only took a month
wayne.cunningham 1/5/2016 2:30:53 AM
Jen-Hsun is highlighting the speed of Nvidia GPUs in analyzing image datasets.
wayne.cunningham 1/5/2016 2:31:42 AM
From simply identifying cars, now the demo shows recognition of pedestrians.
Lori Grunin 1/5/2016 2:32:13 AM
wayne.cunningham 1/5/2016 2:32:30 AM
Now we have five different classes of objects, including cars, pedestrians and traffic signs.
Lori Grunin 1/5/2016 2:33:30 AM
wayne.cunningham 1/5/2016 2:33:39 AM
New demo shows how the computer can recognize each pixel that makes up a person or car
wayne.cunningham 1/5/2016 2:34:19 AM
Here, the car computer can develop a very true image of the world around it
wayne.cunningham 1/5/2016 2:34:59 AM
Demo ramps up the difficulty with new video from Audi, driving in snowy weather on a highway
Lori Grunin 1/5/2016 2:35:01 AM
wayne.cunningham 1/5/2016 2:35:40 AM
The Deep Learning processor was able to identify a car in this imagery that humans had a hard time discerning.
wayne.cunningham 1/5/2016 2:36:11 AM
That was the Drivenet Deep Learning network, which runs on an Nvidia TitanX chip
Lori Grunin 1/5/2016 2:36:23 AM
wayne.cunningham 1/5/2016 2:36:28 AM
Jen-Hsun has the Titan X in his hand how
Lori Grunin 1/5/2016 2:36:52 AM
wayne.cunningham 1/5/2016 2:37:51 AM
Drivenet on Titan X is a reference platform for object recognition and learning, not the part that goes in the car.
Lori Grunin 1/5/2016 2:37:59 AM
wayne.cunningham 1/5/2016 2:38:45 AM
Talking about more Audi development of object recognition using Nvidia technology
Lori Grunin 1/5/2016 2:39:04 AM
wayne.cunningham 1/5/2016 2:39:10 AM
Mercedes-Benz is also using Nvidia tech for Deep Learning object recognition
Lori Grunin 1/5/2016 2:39:39 AM
wayne.cunningham 1/5/2016 2:39:48 AM
ZMP, a Chinese company developing a robo-taxi, is using Nvidia tech, as is BMW
Lori Grunin 1/5/2016 2:39:54 AM
Lori Grunin 1/5/2016 2:40:06 AM
Lori Grunin 1/5/2016 2:40:42 AM