Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


NightHawk NVS

Final Presentations

Luis Carrasco

on 3 May 2010

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of NightHawk NVS

System Summary As a driver’s vision becomes diminished due to darkness and headlight restrictions, the risk of an automobile accident increases. By improving night-time vision and providing alert signals to the driver, the Nighthawk N.V.S system aids drivers in locating obstacles, road signs, lane marks, and possible hazards.

The system consists of an externally mounted front-facing camera, a processing unit that analyzes each frame for potential dangers, and a touch screen user interface system.

This system uses a series of cutting edge image processing and artificial intelligence algorithms to determine whether individual objects in an image represent a threat to the vehicle. System Overview It starts with a car A night vision camera night-time The Requirements 1 The system Will use
a laptop 2 3 4 The system Will use
a touchscreen The system Will use
a night-vision camera The system Will be powered by the car battery 5 7 6 8 11 10 9 12 60 Meter 560x560 resolution 720p actual resolution The system Will alert the driver audio-visually when it detects an object The system Will alert the driver audio-visually when it detects a sign The system will process each frame within .5s actual timing .05s on average The system will detect objects Diamond Shaped Warning Signs Stop Signs Obstacles face > 0.5 squared meters We Do Pedestrians Too ! The car can travel up to 30mph We have tested up to 70mph < 10% error for obstacles <5% error for signs for most cases the error was <5% but the
accuracy of the algorithm depends on occlusion
lighting, speed, etc. Bonus Bonus Bonus Bonus The system can be controlled by the user via touch-screen The system is aware if its surroundings The system can detect lanes The system can detect when the driver is drifting out of a lane The Demo Live Testing The route was chosen to include plenty of road signs and visible lanes (which is difficult in boston) On the live test, the system correctly tagged 18 road signs, it missed one, and did not have any false alarms On the live test, the system correctly identified the driver changing lanes twice, it misdetected twice and did not miss any lane changes. It did have trouble on the right/left turns On the live test, the system correctly tagged 32
obstacles, it missed none, and did not missclassify The surroundings awareness correctly tagged moving objects. However it incorrectly tagged non-moving objects just as much Fun Facts The project took over 1000 hours to complete Not once did we get pulled over by a cop
during testing Known Issues The system performs better at night (it was
designed that way) It does not pick up boxes as well, but it picks
up pedestrians and other obstacles very well. Sign matching can only pick up one of each
kind of sign each frame The system was tested on a fast machine, it will
underperform on other systems The Future The system will be better as a stand alone system Integrated with GPS, the system could serve as an
augmented reality navigation system Once HUDs are affordable, the system could
overlay information on the car's windshield Heat infrared cameras will do a better job of
identifying posible threats on the road The project is made up of more than 2500 lines of code The most difficult part was by far the object detection algorithm. It uses a HOG classifier that took over 96 hours to train A laptop computer vision system The OpenCV Library And a touchscreen in car OpenCV is a computer vision library originally developed by Intel. It is free for use under the open source BSD license. The library is cross-platform. It focuses mainly on real-time image processing. User Interface Camera The System uses a
high definition
night-vision camera Algorithms Power The whole system draws its power from a car battery Project
Overview Lane Detection Sign Detection Surroundings Awareness Object Detection Activated on click/touch Each Algorithm has an on/off switch and a setting to turn the sound alerts on and off The Touch Screen The touchscreen is the main form of user interaction. It has a statusbar that informs the driver the current state of the algorithm. It highlites lanes, signs, and obstacles. It overlays warning messages when a possible threat is encountered. It plays a sound in case of an event. It also houses the user interface, which pops up when the user touches the screen Andrew Sarratori York Chan Sehrish Abid Luis Carrasco Wesley Griswold
Full transcript