TF Classify Image
The TF Classify Image operator is an extension to the base component com.sap.system.processExecutor. It runs tensorflow code that applies the inception neuronal network model (trained on ImageNet 2012 Challenge data set) to recognize the generic content of a picture.
The code is based on the tensorflow Image Recognition example: https://www.tensorflow.org/tutorials/image_recognition
Prerequisites
- Internet connection: Required for downloading the inception model.
Configuration Parameters
|
Parameter |
Type |
Description |
|---|---|---|
|
cmdLine |
string |
The command-line parameter that specifies where the model is
stored in the repository.
Default: "" |
|
modeldir |
string |
Contains the modeldir parameter.
Default: "/modeldir" |
|
skip_frames |
string |
Skips n frames in the classification and considers only every nth frame. Default: "/modeldir" |
|
framebuffer_size |
string |
The number of frames buffered in the operator. Note that incoming frames that would exceed the framebuffer_size are discarded. Default: "/modeldir" |
Input
|
Input |
Type |
Description |
|---|---|---|
|
stdin |
stream.base64.jpg |
The input frames in base64-encoded jpg format. |
Output
|
Output |
Type |
Description |
|---|---|---|
|
stdout |
stream |
The classification results separated by the newline command. |
|
stderr |
stream |
A standard error. |