CogniMem SDK Comparison




  CogniSoft Express
(CogniStix Edition)
CogniSoft Express
(CogniBlox Edition)
CogniSoft Professional
Edition
Windows' drivers X X X
Linux drivers (64-bit)   X X
CogniSoft Data SDK
(supports generic digital data)
X X X
CogniSoft Image SDK
(supports 2-D image digital data)
    X
Arduino support and examples   X X
C, C++ support and examples X X X
C# support and examples X X X
VB.NET support and examples   X X
Java support and examples   X X
MATLAB support and examples     X
Python support and examples     X
CogniSoft Data Trainer     X
CogniSoft Image Trainer     X
CogniSoft CogniCar Demo X    
CogniSoft CogniCoaster Demo X    
CogniSoft CogniScript Demo X    
CogniMem Phone Support     X
CogniMem Email Support X X X

 
 






Back to top...

CogniSoft™ - Data Trainer





The CogniMem Data Trainer is a software application which simplifies the process of "batch" training, recognition, clustering, and validation of CogniMem knowledgebases from pre-defined input pattern files, using a network of CogniMem neurons in either software simulation mode or physical hardware mode. Hardware mode supports both a single CogniBlox module or a stack of CogniBlox modules. The CogniSoft Simulator allows you to define the network capacity from 1,024 neurons (1 CM1K) all the way up to 1,638,400 neurons (400 CogniBlox boards).
 


PRODUCT FEATURES

 

  • Importing user-provided input data patterns from file
  • A random pattern generator
  • Supports single and iterative training modes
  • User-defined maximum iteration count
  • Training logging capabilities
  • Selection of L1 (manhattan) and LMAX norm modes
  • User-defined global Minimum and Maximum Influence values for increased accuracy
  • User-defined "K" value
  • Extensive reporting and analysis of the training statistics (including recognition accuracy metrics)
  • Knowledgebase viewer
  • Knowledgebase importer/exporter
  • Cross-Validation testing
  • A learning curve graph
  • Manual vector training





Back to top...

CogniSoft™ - Image Trainer





The Image Training System software is intended for use with the V1KU module when configured with the CogniSoft™ engine for object recognition.

You can adjust the video acquisition settings, define a region of interest and teach the neurons to recognize and classify this region. Once satisfied with the accuracy of the recognition, you can activate the use of the I/O lines to interface directly with actuators and other applications. The knowledge built by the neurons during the teaching phase can be saved to file for later use and even distribution to other V1KU boards.


PRODUCT FEATURES

 

  • View the live video images to adjust the positioning of the camera
  • Adjust the sensor settings to obtain the best image quality
  • Define the region of interest to monitor
  • Teach up to 1024 categories of objects
  • Monitor a region of interest in live images
  • Record images to disk based on the content of the image
  • Activate the output lines of the camera
  • Save the knowledge you have built by teaching examples