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.


Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks


No description

Oihana Otaegui

on 5 September 2014

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of 1qwer

No real-time detection is feasible
1000+ cameras in large CCTV systems
1-10 human visualization

Post-investigation is required to determine location, camera and time of interest events (thefts, abandoned luggage, accidents, etc)

E.g. Underground station Atocha (Madrid)
100+ cameras
2-4 days for finding a visual evidence of a complex event might involve police investigation, iterative extraction of videos from recorder, etc. +105 researchers (+20 PhD)
International Team
Member of:
IK4 Private non Profit Applied Research Centre in Visual Computing, Knowledge Engineering and Communications Technologies From Industrial Needs To Research Projects From Research Results To Industrial Prototypes / Products R+D in Computer Vision and Knowledge Engineering for Transport sector
Provide advance services and new technology
Improve process effectiveness through intelligent solutions
Clients: private and public sector Infraestructure ADAS Advance Mobility Solutions BackOffice Security ITS and Engineering ROAD
Non Intrusive Vehicle counting and classifying

Computer Vision based track Auscultation Easy Configurable Reat Time Countung, Clasification and Tracking Railway track defects
Early detection
Avoid accidents
Automatic report generation Computer Vision based solution
Easily mounted on commercial trains Track Auscultation Lane Departure Warning (LDW) Collision Avoidance System (CAS) Traffic Sign Detection (TSD) Pedestrian Detection (PD) Driver Attention Detection (DAD) Lane Departure Warning (LDW) Collision Avoidance System (CAS) Traffic Sign Detection (TSD) Pedestrian Detection (PD) Driver Attention Detection (DAD) CCTV Event Detection Complex events: really those of interest (persons fighting, accident, abandoned luggage…)

Humans can do it, but potentially hundred of cameras, thousands of hours of video

Computers can scan all the video archives, but complex events can not be detected with 100% reliability by computer methods Human Computer Hybrid Solution Computer proposes a number of short clips and the human selects
Post-investigation time can be dramatically reduced
The interface can offer numerous options for type of event, filter by date, camera, etc. Multiple objects
Full-body only Person detection & tracking 3D face tracking Facial expressions 3D face model + expression
Haar face detection
Online Appearance Model (OAM)
tracking and fitting
Fit of 3D facial modes Face detection Person detection and tracking Can be used to detect stops, person running… Up to 4-5 persons Performance
Core2 Quad 2.5 GHz
Detection 70 ms
Tracking 4 ms Rao-Blackwellized Data Association Particle Filter (RBDAPF)
Template matching learning Histogram of Oriented Gradients (HOG)
CPU/GPU capabilities Probabilistic tracking through time
+ learning each object Combined detector full-body and head/shoulder at each frame Person detection & tracking Person detection and tracking Viulib® Turning vision into reality Vicomtech-IK4 has combined in Viulib®, a proprietary computer vision technology, consisting of complex algorithms, software and specialized techniques . An solution that collects, processes, and analyzes real-time video images. Viulib® is an advanced software library that - Simplifies the building of complex artificial vision solutions
- Reduces time in prototyping phases
- Facilitates technology validation
- Optimizes scalable deployments
A flexible collaboration framework for your business needs . Making the transfer of high-tech computer vision to the real world quick prototyping facilities for mechanical parts Visual Odometry is a vision based motion estimation algorithm that estimates the ego-motion of a camera through its environment by evaluating the captured images. Driverless Vehicle Sensoring Visual Sensors: Odometry, SLAM SLAM is a technique used by robots and autonomous vehicles to build up a map within an unknown environment , or to update a map within a known environment , while at the same time keeping track of their current location. Low Cost Real Time Motion Sensor Visual EgoMotion 2-cameras - stereo vision EgoMotion estimation and Scena Map building
Full transcript