一名热爱体感技术的
业余专业开发人员

[转]“Kinet for Windows” presented by Michael Fry and John McQueen【Kinect v2.0 台灣首度曝光暨技術研討會】

原文地址

Kinet for Windows 【Kinect v2.0 台灣首度曝光暨技術研討會】
Time: 2014.03.14 2:00 pm – 4:00 pm
Location: 國泰金融大樓國際會議廳(台北市松仁路9號)
Speakers: Michael Fry and John McQueen (New Business Development, Kinet for Windows)
Agenda:
-What is NUI and K4W?
-Industry Scenarios
-Kinect V2 sensor (Differences from V1)
– Kinect V2 Data Sources
– Kinect V2 Helper Objects

Kinet for Windows

What is NUI?
Natural User Interaction (NUI) is a completely new way to experience technology. It is a paradigm shift that allows us to go beyond only manual input to communicating with our technology in a more natural way.

“For many years now, Microsoft has been working on NUI-based technologies such as speech, touch, contextual and environmental awareness, immersive 3D experiences, and anticipatory computing; all with the goal of a computer that can see, listen, learn, talk and act smartly on our behalf.” – Steve Ballmer

NUI is the most powerful and most “natural” when it is multi modal.

K4W What is Kinet for Windows?
Our mission is to evolve Kinect.
Hardware + Software
K4W drives value in retail
Kinect, Next Generation
USB 3.0, much more data!
1080p video
3X precision of depth data
Better audio
Wider field of view

Sensor differences from V1 (briefly, FOV)
FOV Old:
53×47
Tilt Motor to handle
~2 meters away to see a user

FOV New:
70×60
No tilt motor (wider field of view)
~1 meter away to see a user

Sensor differences from V1 (Infrared)
Old:
Color or Infrared, not both
Speckled pattern on Infrared (structured light)

New:
Color and infrared
No speckled pattern (time of flight)
Clean infrared frames (Ambient light subtraction)

Sensor differences from V1 (briefly… Color)
Color (Old): 640×480, 30 Hz or 1280×960, 12 Hz or; 640×480 Infrared, 30 Hz
Color (New):
One color camera resolution (1920×1080), frame rate (30 Hz*)
One depth/IR camera resolution (512×424), frame rate (30 Hz)
Clean infrared frames
Can use infrared and color cameras simultaneously

Data sources
SENSOR: Depth, Color, Infrared
DERIVED: Body index, Body, Audio

Depth
– 3x the fidelity of V1
– Wider FOV
– .5 meters to 4.5 meters (no “near mode” switch required)
– Frame data is 2 bytes per pixel: 16-bit distance millimeters

Body
– Formerly known (in V1) as Skeleton
– Frame data is an array of Body objects
– Many new features in V2
– More joits (neck, thumbs, hand tips)
– Hand states (open, closed, “lasso”)
Activities (eye closed, mouth open, mouth motion, looking away)
– Appearance (wearing glasses)

“V1 is awesome for gross human motions while V2 with incredible details and more data, is great for precise motions.”

V2 will soon to be released around mid summer this year in US and a bit later than that for Taiwan.

Check out a professional report in Chinese about this event 超專業的中文記錄請看:《Kinect v2.0 台灣首度曝光暨技術研​討會》參加紀錄