CIB image toolbox technical manual
12. CIB aiModule
Properties
Description
Currently, CIB aiModule is only called via CIB image toolbox. Its functionality is to detect boundaries in order to auto crop images correctly.
The following libraries are required to run CIB aiModule:
|
Version |
Library |
|
Windows 64 |
CibAiModule64.dll, tensorflow.dll |
|
Linux x86 64 |
libcibaimodule64.so, libgcc_s.so.1, libstdc++.so.6, libtensorflow.so, libtensorflow_framwork.so |
|
Android x86 64 |
libcibaimodule64.so |
|
Android arm 64 |
libcibaimodule64.so |
Properties
The following properties are required for CIB image toolbox to call cropping with CIB aiModule:
|
Property name |
Description |
Default value |
|
OperationName |
Name |
InfoGetBoundaries |
|
Profile:InfoGetBoundaries/Algorithm |
Which algorithm to use for border detection: New, Original, OriginalReduced, Ai. Ai requires the Ai module. |
Ai |
|
Profile:InfoGetBoundaries/ModelFilename |
Path to the file that contains trained data for requested operation (now only image-cropping model exists) |
<Path_To_Model_File> |
|
Profile:InfoGetBoundaries/ConfidenceIncluded |
|
0 |
Further supported input properties of CIB aiModule:
|
Property name |
Description |
Default value |
|
InputMemoryAddress |
string representation of an address in the memory where input image data is provided (image data as an in-memory Bitmap) |
- |
|
Width |
Width of the input image, provided by InputMemoryAddress |
- |
|
Height |
Height of the input image, provided by InputMemoryAddress |
- |
Output properties of CIB aiModule:
|
Property name |
Description |
Default value |
|
Result |
Contains the result. For image-cropping, the result contains XY-coordinates of four vertices (8 numbers) in string format, numbers are separated by semicolon |
- |