Missed TensorFlow World? Check out the recap. Learn more

tensorflow::serving::ClassifierInterface

This is an abstract class.

#include <classifier.h>

Model-type agnostic interface for performing classification.

Summary

Specific implementations will exist for different model types (e.g. TensorFlow SavedModel) that can convert the request into a model specific input and know how to convert the output into a generic ClassificationResult.

Constructors and Destructors

~ClassifierInterface()

Public functions

Classify(const ClassificationRequest & request, ClassificationResult *result)=0
virtual Status
Given a ClassificationRequest, populates the ClassificationResult with the result.

Public functions

Classify

virtual Status Classify(
  const ClassificationRequest & request,
  ClassificationResult *result
)=0

Given a ClassificationRequest, populates the ClassificationResult with the result.

Details
Parameters
request
Input request specifying the model/signature to query along with the data payload.
result
The output classifications that will get populated.
Returns
A status object indicating success or failure.

~ClassifierInterface

virtual  ~ClassifierInterface()=default