Applications



01. Applications

In addition to the above cases, it is possible to function as a core part of an artificial intelligence sensor in a system of various fields depending on the user.


02. CCTV Surveillance System

Presence Detection

When a specific object appears on the camera, it immediately sounds an alarm or notifies you via the web interface.

For examples,

  • Alert when a person appears in the security area at night
  • Alert when a harmful animal appears in the farm,
  • Alert if a fire occurs

Absence Detection

If a specific object does not appear in the camera for a specified time, it will notify you via the web interface.

For examples,

  • Alert if the security personnel has an accident and is not visible for a specified time
  • Alert if an elderly living alone does not show up for a while
  • Immediate alarm when a valuable on exhibit disappears

Counting

Recognize and record the number of people or objects passing through the camera view.

  • You can collect information on how many people are in each corner of your store over time.
  • Identify the number of people entering or exiting through a specific door (or designated area) by time
  • At a car intersection, you can see how many cars are moving in each direction, instantaneously or by time of day.

License Plate Recognition

  • The Deep Runner can be configured to combine two deep learning models. For example, the first model recognizes the number plate in the car and the second model recognizes the numbers in the number plate image. Therefore, it can recognize the license plate from car images.
  • In a similar way, it can read the serial number from products.
  • It might also recognize that someone on the street is wearing a hat or carrying a bag.

Moving Direction Detection

  • It detects the direction of moving objects.


03. Automated Quality Control System


Build your own automated quality inspection system with Deep Runner


  • Reduce costs with automation
  • Can be applied to any product by training the system
  • Support for the latest deep learning algorithms
  • PLC interworking support

The Deep Runner is a core component for building automated defect product detection systems


  • Support for deep recognition algorithms with the highest recognition rate: Xception, MobileNet, GoogleNet
  • Train with product images to recognize defect products using easy-to-use Deep Trainer Software
  • Split recognition function to detect small defects in high quality product image
  • Pinout recognition results and capture command using PLC interworking function
  • Capture function to automatically acquire training product image

Opportunity to build highly value-added factory automation systems at low cost

04. Recognition result output


  • Video monitor output
  • Ethernet port - Recognition result record real-time transmission
  • GPIO - Serial port, PIN OUT


05. Training Procedure

  • Deep learning software 'Deep Trainer' provides free at the time of purchase
  • Provides the ability to recognize 90 objects for security systems.
  • User-specific objects can be trained by using images
  • The user needs to acquire a lot of sample images necessary for learning about the recognition object.


06. Who can use it?

  • System Developer
  • Domestic and overseas entrepreneurs who realize new ideas
  • DIY developers such as robots
  • Anyone who wants to make deep learning training easier





For more information, see Deep Runner User Manual 13. Application Scenarios.