AnomalyDetection.h¶

Anomaly Detection is a function that learns only normal images and detects abnormal images and their areas.

SAIGE VISION provides the Anomaly Detection function that detects abnormal images and their areas by learning only normal images. The Anomaly Detection function detects the abnormal area as a SaigeSegmentedObject structure, which is a Contour. Therefore, it is similar to the Segmentation function, and additionally provides whether the image is abnormal, and the overall abnormal score of the image.

Inference Options¶

You can set whether to calculate the Mask, Object, and Heatmap, which are the inference results of Anomaly Detection. The Scoremap is calculated by default and cannot be set not to be calculated.

// ExampleIADRuntime.cpp:155

options->calcScoremap = true;
options->calcHeatmap = true;
options->calcMask = true;
options->calcObject = true;
options->calcObjectAreaAndApplyThreshold = true;
options->calcObjectScoreAndApplyThreshold = true;
unsigned int flags = SAIGE_SET_CALC_SCOREMAP | SAIGE_SET_CALC_HEATMAP |
                     SAIGE_SET_CALC_MASK | SAIGE_SET_CALC_OBJECT |
                     SAIGE_SET_CALC_OBJECT_AREA_AND_APPLY_THRESHOLD |
                     SAIGE_SET_CALC_OBJECT_SCORE_AND_APPLY_THRESHOLD;
retval = SaigeSetInferenceOption(handle, options, flags);
if (retval < 0) {
    std::cerr << "SaigeSetInferenceOption() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}

See also:

Heatmap¶

In Anomaly Detection, the Scoremap can be colored as a Heatmap. The Score in Anomaly Detection is expressed as a real number greater than or equal to 0, unlike other functions. Therefore, you can set the maximum and minimum Score values corresponding to the color area. The trained model has a suitable Score range calculated and set, but you can adjust the range if it is not clearly visible.

// ExampleIADRuntime.cpp:194

options->heatmapColorRangeMin = 1.0;
options->heatmapColorRangeMax = 2.5;
retval = SaigeSetInferenceOption(handle, options, SAIGE_SET_HEATMAP_COLOR_RANGE);
if (retval < 0) {
    std::cerr << "SaigeSetInferenceOption() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}

Before

After

Inference Results¶

You can get the basic inference results of Anomaly Detection, isNg, anomalyScore, scoreMin, and scoreMax.

// ExampleIADRuntime.cpp:300

// DEFAULT
retval = SaigeGetInferenceResultAnomalyScores(result, &anomalyScore, &scoreMin, &scoreMax);
if (retval < 0) {
    std::cerr << "SaigeGetInferenceResultAnomalyScores() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}
retval = SaigeGetInferenceResultIsNg(result, &isNg);
if (retval < 0) {
    std::cerr << "SaigeGetInferenceResultIsNg() failed - " << SaigeStrError(retval)
              << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}

ScoremapIAD¶

The Scoremap, which is the inference result of Anomaly Detection, is represented as a float array to express a real number greater than or equal to 0.

// ExampleIADRuntime.cpp:345

retval = SaigeGetInferenceResultScoremapIAD(result, &scoremap);
if (retval < 0) {
    std::cerr << "SaigeGetInferenceResultScoremapIAD() failed - "
              << SaigeStrError(retval) << std::endl;
    SaigeDestroyInferenceHandle(handle);
    SaigeDestroyInferenceResult(result);
    return retval;
}
if (verbosity >= 4) {
    SaigeShowImage(scoremap, w, h, 1, SaigeImageType::SV_F32, false, false,
                   1 / scoreMax, waitkeyMilliSecond);
}

For other results such as Mask, Object, Heatmap, please refer to the example code and API documentation.