Classification.h¶

Classification classifies images into several classes determined by the user.

The image is classified into the class with the highest score, which is calculated as the probability of belonging to each class.

Inference Options¶

See also: SaigeInferenceOption

Class Activation Map¶

In the Classification function, you can calculate the Class Activation Map. This provides distribution information on the degree of class activation for each pixel in the image. To use this feature, set the calcClassActivationMap value to true and set the flag to SAIGE_SET_CALC_CLASS_ACTIVATION_MAP.

Warning

It cannot be calculated when the optimization feature is applied. In this example, the Class Activation Map is set to be calculated only when the optimize option value is false.

// ExampleClsRuntime.cpp:166

if (verbosity >= 3 && optimize == false) {
    options->calcClassActivationMap = true;
    retval = SaigeSetInferenceOption(handle, options, SAIGE_SET_CALC_CLASS_ACTIVATION_MAP);
    if (retval < 0) {
        std::cerr << "SaigeSetInferenceOption() failed - " << SaigeStrError(retval)
                  << std::endl;
        SaigeDestroyInferenceHandle(handle);
        SaigeDestroyInferenceResult(result);
        return retval;
    }
}

Inspection Results¶

In the Classification function, you can get the Score values per class, the class ID of the classified image, and the inference time.

Additional Scores¶

You can set the additional Scores per class, SaigeInferenceOption::additionalScores. This allows you to give weight to specific classes. To get the result with this option applied, you must pass true as the second argument to the SaigeGetInferenceResultScore function.

// ExampleClsRuntime.cpp:246

std::vector<float> score;
int classId;
score.resize(modelInfo->numOfClasses);
retval = SaigeGetInferenceResultScore(result, score.data(), true);
if (retval < 0) {
    std::cerr << "SaigeGetInferenceResultScore() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}

retval = SaigeGetInferenceResultClass(result, &classId);
if (retval < 0) {
    std::cerr << "SaigeGetInferenceResultClass() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}

retval = SaigeGetInferenceResultInspectionTime(result, &imreadTime, &inferenceTime,
                                               &postProcessingTime);
if (retval < 0) {
    std::cerr << "SaigeGetInferenceResultClass() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}