SaigeVisionAPI.h¶

#include <stdint.h>
#include "saige_error.h"

SaigeVision 추론 기능에 필요한 핵심적인 함수, 상수, 구조체, 플래그 등이 정의되어 있습니다.

Classes¶

Type	Name
struct	SaigeHandleConfig Structure containing configuration settings for the inference handle, which can impact GPU memory and inference time.
struct	SaigeRoiInfo Structure containing ROI information. Composed of the ROI mode and corresponding parameters, along with common blind mask information.
struct	SaigeModelInfo Structure containing model information.
struct	SaigeTextFormatting (OCR) Option for specifying the format of text to be recognized. There are two approaches: setting a character set for recognition regardless of character count, or defining character sets based on character position.
struct	SaigeInferenceOption inference options struct
struct	SaigeRotatedBox Structure containing information about a rotatable rectangle.
struct	SaigeDetectedObject Structure containing information about a detected area from Object Detection (DET).
struct	SaigeEllipse Structure containing information about an ellipse.
struct	SaigeSegmentedObject Structure containing information about a detected area from Segmentation (SEG) or Anomaly Detection (IAD).
struct	SaigeTextObject SaigeTextObject.

Public Types¶

Type	Name
typedef void *	SaigeInferenceHandle A handle used to manage information and perform inference.
typedef void *	SaigeInferenceResult A handle used to manage inference results.

Public Attributes¶

Type	Name
constexpr int	kProjectTypeCLS = = 0 Constant value indicating Classification project.
constexpr int	kProjectTypeDET = = 1 Constant value indicating Detection project.
constexpr int	kProjectTypeSEG = = 2 Constant value indicating Segmentation project.
constexpr int	kProjectTypeIAD = = 3 Constant value indicating Anomaly Detection project.
constexpr int	kProjectTypeOCR = = 4 Constant value indicating OCR project.
constexpr int	kProjectTypeIEN = = 5 Constant value indicating IEN project.
constexpr int	kMaxLenOfClassName = = 63 Constant value indicating the max length of class name.
constexpr int	kMaxNumOfClasses = = 255 Maximum number of classes.
constexpr int	kMaxLenOfModelName = = 63 Constant value indicating the max length of model name.
constexpr int	kMaxLenOfCameraName = = 63 Constant value indicating the max length of camera name.
constexpr int	kMaxLenOfTrainedVersion = = 63 Constant value indicating the max length of version information string.
constexpr int	kMaxLenOfToolName = = 63 kMaxLenOfToolName
constexpr int	kMaxLenOfTextFormatting = = 200 (OCR) Maximum length of recognized text. A recognized text in a single SaigeTextObject cannot exceed this length.

Public Functions¶

Type	Name
int	SaigeInferFromImageFile (SaigeInferenceHandle handle, const char * imagePath, SaigeInferenceResult result) Performs inference using the specified image file path.
int	SaigeInferFromByteArray (SaigeInferenceHandle handle, const unsigned char * imageData, const int width, const int height, const unsigned int flags, SaigeInferenceResult result) Performs inference on a single image data array. For 3-channel color images, RGB ordering is preferred for speed in real-time applications. If images are acquired in RGB, it is recommended to use RGB.
int	SaigeInferFromByteArrayWithProductId (SaigeInferenceHandle handle, const unsigned char * imageData, const int width, const int height, const unsigned int flags, const char * productId, SaigeInferenceResult result) Performs inference on a single image data array with an additional product identifier. This is similar to SaigeInferFromByteArray, with the addition of a product ID, which is a user-defined identifier for the inspection target.
int	SaigeInferFromMultipleByteArray (SaigeInferenceHandle handle, const int numOfImages, const unsigned char imageData, const int width, const int height, const unsigned int flags, SaigeInferenceResult** * results) Performs batch inference on multiple images of the same size.
int	SaigeInferFromMultiPageByteArray (SaigeInferenceHandle handle, const int numOfPages, const unsigned char ** imageData, const int width, const int height, const int * channels, const unsigned int flags, SaigeInferenceResult result) Performs inference using multiple pages of image data.
int	SaigeGetInferenceResultScore (SaigeInferenceResult result, float * scores, bool includeAddScore) (CLS) Retrieves an array of scores per class.
int	SaigeGetInferenceResultClass (SaigeInferenceResult result, int * classIndex) (CLS) Retrieves the index of the class with the highest final score, including additional score.
int	SaigeGetInferenceResultClassActivationMap (SaigeInferenceResult result, unsigned char ** classActivationMap) (CLS) Retrieves the class activation map array.
int	SaigeGetInferenceResultNumOfDetectedObjects (SaigeInferenceResult result, int * numOfObjects) (DET) Retrieves the number of detected objects from Object Detection.
int	SaigeGetInferenceResultDetectedObjects (SaigeInferenceResult result, SaigeDetectedObject * detectedObjects) (DET) Retrieves an array of detected objects from Object Detection.
int	SaigeGetInferenceResultNumOfSegmentedObjects (SaigeInferenceResult result, int * num) Retrieves the number of segmented objects detected by Segmentation (SEG) or Anomaly Detection (IAD).
int	SaigeGetInferenceResultSegmentedObjects (SaigeInferenceResult result, SaigeSegmentedObject * segObjects) Retrieves an array of segmented objects detected by Segmentation (SEG) or Anomaly Detection (IAD).
int	SaigeGetInferenceResultMask (SaigeInferenceResult result, unsigned char ** mask) Retrieves a Mask array from Segmentation (SEG) or Anomaly Detection (IAD).
int	SaigeGetInferenceResultScoremapSEG (SaigeInferenceResult result, unsigned char ** scoremap) Retrieves the Scoremap array for Segmentation (SEG).
int	SaigeGetInferenceResultIsOversized (SaigeInferenceResult result, bool * isOversized) Checks if the input image size exceeded the inspection size set in `SaigeHandleConfig`.
int	SaigeGetInferenceResultAnomalyScores (SaigeInferenceResult result, double * anomalyScore, double * minScore, double * maxScore) Retrieves three types of scores for Anomaly Detection (IAD).
int	SaigeGetInferenceResultIsNg (SaigeInferenceResult result, bool * isNg) Retrieves the normal/abnormal classification result for the image in Anomaly Detection (IAD).
int	SaigeGetInferenceResultScoremapIAD (SaigeInferenceResult result, float ** scoremap) Retrieves the Scoremap array for Anomaly Detection (IAD).
int	SaigeGetInferenceResultHeatmap (SaigeInferenceResult result, float ** heatmap) SaigeGetInferenceResultHeatmap.
int	SaigeGetInferenceResultNumOfTextObjects (SaigeInferenceResult result, int * num) SaigeGetInferenceResultNumOfTextObjects.
int	SaigeGetInferenceResultTextObjects (SaigeInferenceResult result, SaigeTextObject * textObjects) SaigeGetInferenceResultTextObjects.
int	SaigeGetInferenceResultEnhancedImage (SaigeInferenceResult result, unsigned char ** image) SaigeGetInferenceResultEnhancedImage.
int	SaigeGetInferenceResultInspectionTime (SaigeInferenceResult result, long long * imreadTime, long long * inferenceTime, long long * postProcessingTime) SaigeGetInferenceResultInspectionTime.
int	SaigeGetInferenceResultProductId (SaigeInferenceResult result, const char ** productId, unsigned int * length) SaigeGetInferenceResultProductId.
int	SaigeInitializeAlignModule (const int gpuIndex, const int sourceWidth, const int sourceHeight, const int targetWidth, const int targetHeight) Function to activate the Saige Alignment function.
int	SaigeFinalizeAlignModule () Function to terminate the Saige Alignment function. Releases the GPU memory used by the function.
int	SaigeAlignImageFromByteArray (const unsigned char * sourceImage, const int sourceWidth, const int sourceHeight, const int sourceChannel, const unsigned char * targetImage, const int targetWidht, const int targetHeight, const int targetChannel, unsigned int flag, unsigned char * croppedSourceImage) Function to adjust the position of the Source image to match the Target image.
int	SaigeAlignImageFromFile (const char * sourceImagePath, const char * targetImagePath, int * cropWidth, int * cropHeight, int * cropChannel, unsigned char * croppedSourceImage) Not Implemented.
int	SaigeCreateInferenceHandle (SaigeInferenceHandle * handle, const int projectType, const char * modelFilePath, const int deviceIndex, SaigeHandleConfig * config, const unsigned int flags) Creates an inference handle.
int	SaigeDestroyInferenceHandle (SaigeInferenceHandle handle) Destroys an inference handle.
int	SaigeCreateInferenceResult (SaigeInferenceResult * result) Creates a result handle.
int	SaigeDestroyInferenceResult (SaigeInferenceResult result) Destroys a result handle.
int	SaigeGetModelInfo (SaigeInferenceHandle handle, SaigeModelInfo * info) Retrieves model information from the inference handle.
int	SaigeSetInferenceOption (SaigeInferenceHandle handle, SaigeInferenceOption * option, const unsigned int flags) Sets inference option values on the inference handle.
int	SaigeGetInferenceOption (SaigeInferenceHandle handle, SaigeInferenceOption * option) Retrieves inference option values from the inference handle.
const char *	SaigeStrError (const int errnum) SaigeStrError.
const char *	SaigeGetModuleVersion () SaigeGetModuleVersion.
int	SaigeSetHandleMetadata (SaigeInferenceHandle handle, const char * key, const char * value) SaigeSetHandleMetadata.
int	SaigeSetInferenceMetadata (SaigeInferenceResult result, const char * key, const char * value) SaigeSetInferenceMetadata.
int	SaigeSetResultMetadata (SaigeInferenceResult result, const char * key, const char * value) SaigeSetResultMetadata.

Macros¶

Type	Name
define	SAIGE_WRITE_ON_MODEL_FILE 0x01 (Common) When setting inference option values using SaigeSetInferenceOption, this flag saves the option values to the model file as well.
define	SAIGE_SET_CALC_TIME 0x2 (Common) Flag to set the `calcTime` option.
define	SAIGE_SET_OVERSIZED_IMAGE_HANDLING 0x4 (SEG, OCR, IEN) Flag to set the `oversizedImageHandling` option.
define	SAIGE_SET_ADDITIONAL_SCORES 0x8 (CLS) Flag to set the `additionalScores` option.
define	SAIGE_SET_CALC_CLASS_ACTIVATION_MAP 0x10 (CLS) Flag to set the `calcClassActivationMap` option.
define	SAIGE_SET_MAX_NUM_OF_DETECTED_OBJECTS 0x20 (DET) Flag to set the `maxNumOfDetectedObjects` option.
define	SAIGE_SET_CALC_SCOREMAP 0x40 (SEG) Flag to set the `calcScoremap` option.
define	SAIGE_SET_CALC_OBJECT 0x80 (SEG, IAD) Flag to set the `calcObject` option.
define	SAIGE_SET_CALC_OBJECT_AREA_AND_APPLY_THRESHOLD 0x100 (SEG, IAD) Flag to set the `calcObjectAreaAndApplyThreshold` option.
define	SAIGE_SET_CALC_OBJECT_SCORE_AND_APPLY_THRESHOLD 0x200 (SEG, IAD) Flag to set the `calcObjectScoreAndApplyThreshold` option.
define	SAIGE_SET_OBJECT_BOUNDARY_THRESHOLD 0x400 (SEG, IAD) Flag to set `objectBoundaryThresholdPerClass` in SEG and `objectBoundaryThreshold` in IAD.
define	SAIGE_SET_OBJECT_AREA_THRESHOLD 0x800 (SEG, DET, IAD) Flag to set `objectAreaThresholdPerClass` in SEG and DET, and `objectAreaThreshold` in IAD.
define	SAIGE_SET_OBJECT_SCORE_THRESHOLD 0x1000 (SEG, DET, IAD) Flag to set `objectScoreThresholdPerClass` in SEG and DET, and `objectScoreThreshold` in IAD.
define	SAIGE_SET_CALC_MASK 0x2000 (SEG, IAD) Flag to set the `calcMask` option.
define	SAIGE_SET_CALC_HEATMAP 0x4000 (IAD) Flag to set the `calcHeatmap` option.
define	SAIGE_SET_ANOMALY_SCORE_THRESHOLD 0x8000 (IAD) Flag to set the `anomalyScoreThreshold` option.
define	SAIGE_SET_HEATMAP_COLOR_RANGE 0x10000 (IAD) Flag to set `heatmapColorRangeMin` and `heatmapColorRangeMax` options.
define	SAIGE_SET_POLYGON_SCORE_THRESHOLD 0x20000 (OCR) Flag to set the `polygonScoreThreshold` option.
define	SAIGE_SET_TEXT_SIZE_RANGE 0x40000 (OCR) Flag to set `textSizeRangeMin` and `textSizeRangeMax` options.
define	SAIGE_SET_NUM_OF_TEXT_OBJECTS 0x80000 (OCR) Flag to set the `maxNumOfTextObjects` option.
define	SAIGE_SET_TEXT_SCORE_THRESHOLD 0x100000 (OCR) Flag to set the `textScoreThreshold` option.
define	SAIGE_SET_TEXT_FORMATTING 0x200000 (OCR) Flag to set the `textFormatting` option.
define	SAIGE_IMAGE_COLOR 0x01 Flag indicating color image data.
define	SAIGE_IMAGE_16BIT 0x02 Flag indicating 16-bit per channel image data.
define	SAIGE_IMAGE_ALPHA 0x04 Flag indicating data with an alpha channel.
define	SAIGE_IMAGE_BGR 0x08 Flag indicating the color channel order is Blue, Green, Red.
define	SAIGE_IMAGE_8BIT_GRAY 0x00 Macro indicating 8-bit grayscale image data. It is a single-channel array with a size of Width x Height.
define	SAIGE_IMAGE_24BIT_RGB (SAIGE_IMAGE_COLOR) Macro indicating 24-bit RGB color image data. It is an array of size Width x Height x Channels (3) with channels ordered as R, G, B.
define	SAIGE_IMAGE_24BIT_BGR (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_BGR) Macro indicating 24-bit BGR color image data. It is an array of size Width x Height x Channels (3) with channels ordered as B, G, R.
define	SAIGE_IMAGE_32BIT_RGBA (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_ALPHA) Macro indicating 32-bit RGBA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as R, G, B, A.
define	SAIGE_IMAGE_32BIT_BGRA (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_ALPHA \| SAIGE_IMAGE_BGR) Macro indicating 32-bit BGRA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as B, G, R, A.
define	SAIGE_IMAGE_16BIT_GRAY (SAIGE_IMAGE_16BIT) Macro indicating 16-bit grayscale image data. It is a single-channel array of size Width x Height, with each pixel represented as an unsigned short.
define	SAIGE_IMAGE_48BIT_RGB (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_16BIT) Macro indicating 48-bit RGB color image data. It is an array of size Width x Height x Channels (3) with channels ordered as R, G, B, and each pixel represented as an unsigned short.
define	SAIGE_IMAGE_48BIT_BGR (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_16BIT \| SAIGE_IMAGE_BGR) Macro indicating 48-bit BGR color image data. It is an array of size Width x Height x Channels (3) with channels ordered as B, G, R, and each pixel represented as an unsigned short.
define	SAIGE_IMAGE_64BIT_RGBA (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_16BIT \| SAIGE_IMAGE_ALPHA) Macro indicating 64-bit RGBA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as R, G, B, A, and each pixel represented as an unsigned short.
define	SAIGE_IMAGE_64BIT_BGRA (SAIGE_IMAGE_COLOR \| SAIGE_IMAGE_16BIT \| SAIGE_IMAGE_ALPHA \| SAIGE_IMAGE_BGR) Macro indicating 64-bit BGRA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as B, G, R, A, and each pixel represented as an unsigned short.
define	SAIGEVISION_RUNTIME_VERSION_MAJOR 2 Major version of the SAIGE VISION Runtime C/C++ SDK.
define	SAIGEVISION_RUNTIME_VERSION_MINOR 12 Minor version of the SAIGE VISION Runtime C/C++ SDK.
define	SAIGEVISION_RUNTIME_VERSION_PATCH 3 Patch version of the SAIGE VISION Runtime C/C++ SDK.
define	SV_STR (x) #x
define	SV_VER (M, m, p) SV_STR(M) "." SV_STR(m) "." SV_STR(p)
define	SAIGEVISION_RUNTIME_VERSION "2.12.3" The version of the SaigeVision Runtime C/C++ SDK (deprecated). Please use MAJOR, MINOR, and PATCH versions instead.
define	SAIGE_SET_BATCH_SIZE 0x01 (CLS, DET, SEG, IAD) Flag to set batch size for inference. Used when batch inference API is applied.
define	SAIGE_SET_INSPECTION_SIZE 0x02 (SEG, OCR) Flag to set the width and height of images for inference.
define	SAIGE_OPTIMIZE_HANDLE 0x04 (Common) Flag to apply optimization to the inference handle. Applying inference optimization reduces inference time but may slightly alter results.
define	SAIGE_DO_NOT_INSPECT "do_not_inspect" (SEG) An option for `oversizedImageHandling`, where inference is not performed and an error is raised.
define	SAIGE_DO_NOT_ENHANCE "do_not_enhance" (IEN) An option for `oversizedImageHandling`, where inference is not performed and an error is raised.
define	SAIGE_RESIZE_TO_FIT "resize_to_fit" (SEG) An option for `oversizedImageHandling`. When set, the image is resized to fit within the inspection size, maintaining its aspect ratio, and then inference is performed.
define	SAIGE_CROP_INTO_TILES "crop_into_tiles" (SEG, IEN) An option for `oversizedImageHandling`. When set, the image is cropped into tiles based on the inspection size. Inference is performed on each tile, and results are combined.

Public Types Documentation¶

`SaigeInferenceHandle`¶

A handle used to manage information and perform inference.

void * SaigeInferenceHandle;

`SaigeInferenceResult`¶

A handle used to manage inference results.

void * SaigeInferenceResult;

Public Attributes Documentation¶

`kProjectTypeCLS`¶

Constant value indicating Classification project.

constexpr int kProjectTypeCLS;

`kProjectTypeDET`¶

Constant value indicating Detection project.

constexpr int kProjectTypeDET;

`kProjectTypeSEG`¶

Constant value indicating Segmentation project.

constexpr int kProjectTypeSEG;

`kProjectTypeIAD`¶

Constant value indicating Anomaly Detection project.

constexpr int kProjectTypeIAD;

`kProjectTypeOCR`¶

Constant value indicating OCR project.

constexpr int kProjectTypeOCR;

`kProjectTypeIEN`¶

Constant value indicating IEN project.

constexpr int kProjectTypeIEN;

`kMaxLenOfClassName`¶

Constant value indicating the max length of class name.

constexpr int kMaxLenOfClassName;

`kMaxNumOfClasses`¶

Maximum number of classes.

constexpr int kMaxNumOfClasses;

`kMaxLenOfModelName`¶

Constant value indicating the max length of model name.

constexpr int kMaxLenOfModelName;

`kMaxLenOfCameraName`¶

Constant value indicating the max length of camera name.

constexpr int kMaxLenOfCameraName;

`kMaxLenOfTrainedVersion`¶

Constant value indicating the max length of version information string.

constexpr int kMaxLenOfTrainedVersion;

`kMaxLenOfToolName`¶

kMaxLenOfToolName

constexpr int kMaxLenOfToolName;

`kMaxLenOfTextFormatting`¶

(OCR) Maximum length of recognized text. A recognized text in a single SaigeTextObject cannot exceed this length.

constexpr int kMaxLenOfTextFormatting;

Public Functions Documentation¶

`SaigeInferFromImageFile`¶

Performs inference using the specified image file path.

int SaigeInferFromImageFile (
    SaigeInferenceHandle handle,
    const char * imagePath,
    SaigeInferenceResult result
)

Parameters:

handle The inference handle to use for inference.
imagePath Path to the image file for inference. Supported extensions are jpg, jpeg, png, ppm, bmp, pgm, tif, tiff, and webp.
result The result handle to store the inference results.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeInferFromByteArray`¶

Performs inference on a single image data array. For 3-channel color images, RGB ordering is preferred for speed in real-time applications. If images are acquired in RGB, it is recommended to use RGB.

int SaigeInferFromByteArray (
    SaigeInferenceHandle handle,
    const unsigned char * imageData,
    const int width,
    const int height,
    const unsigned int flags,
    SaigeInferenceResult result
)

Parameters:

handle The inference handle to use.
imageData The image data to be used for inference.
width Width of the image data.
height Height of the image data.
flags Flags to set the color and bit depth of the image.
result The result handle to store inference results.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeInferFromByteArrayWithProductId`¶

Performs inference on a single image data array with an additional product identifier. This is similar to SaigeInferFromByteArray, with the addition of a product ID, which is a user-defined identifier for the inspection target.

int SaigeInferFromByteArrayWithProductId (
    SaigeInferenceHandle handle,
    const unsigned char * imageData,
    const int width,
    const int height,
    const unsigned int flags,
    const char * productId,
    SaigeInferenceResult result
)

Parameters:

handle The inference handle to use.
imageData The image data to be used for inference.
width Width of the image data.
height Height of the image data.
flags Flags to set the color and bit depth of the image.
productId The product ID for the item to be inferred.
result The result handle to store inference results.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeInferFromMultipleByteArray`¶

Performs batch inference on multiple images of the same size.

int SaigeInferFromMultipleByteArray (
    SaigeInferenceHandle handle,
    const int numOfImages,
    const unsigned char ** imageData,
    const int width,
    const int height,
    const unsigned int flags,
    SaigeInferenceResult * results
)

Parameters:

handle The inference handle to use.
numOfImages Number of images in the data.
imageData Array of image data for inference.
width Width of the image data.
height Height of the image data.
flags Flags to set the color and bit depth of the image.
results Array of result handles to store inference results.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeInferFromMultiPageByteArray`¶

Performs inference using multiple pages of image data.

int SaigeInferFromMultiPageByteArray (
    SaigeInferenceHandle handle,
    const int numOfPages,
    const unsigned char ** imageData,
    const int width,
    const int height,
    const int * channels,
    const unsigned int flags,
    SaigeInferenceResult result
)

Parameters:

handle The inference handle to use.
numOfPages Number of pages of image data.
imageData Array of image data for inference.
width Width of the image data.
height Height of the image data.
channels Array of the number of channels in the image data. Each element represents the number of channels for each page.
flags Flags to set the color and bit depth of the image.
result The result handle to store inference results.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultScore`¶

(CLS) Retrieves an array of scores per class.

int SaigeGetInferenceResultScore (
    SaigeInferenceResult result,
    float * scores,
    bool includeAddScore
)

Parameters:

result The result handle containing inference results.
scores [out] Array containing scores for each class.
includeAddScore If set to true, retrieves scores including additional scores. if false, retrieves original scores only.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultClass`¶

(CLS) Retrieves the index of the class with the highest final score, including additional score.

int SaigeGetInferenceResultClass (
    SaigeInferenceResult result,
    int * classIndex
)

Parameters:

result The result handle containing inference results.
classIndex [out] Index of the class with the highest score.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultClassActivationMap`¶

(CLS) Retrieves the class activation map array.

int SaigeGetInferenceResultClassActivationMap (
    SaigeInferenceResult result,
    unsigned char ** classActivationMap
)

The class activation map is a heatmap that shows which image areas contributed most to a specific classification.
Pixel values range from 0 to 255, with higher values indicating a greater contribution to the classification result.
Shape is (original image height H, original image width W).
Internally, this is represented as a 1D array using row-major order. The value at row i, column j is located at data[W*i + j].

Parameters:

result The result handle containing inference results.
classActivationMap [out] Array containing the class activation map.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultNumOfDetectedObjects`¶

(DET) Retrieves the number of detected objects from Object Detection.

int SaigeGetInferenceResultNumOfDetectedObjects (
    SaigeInferenceResult result,
    int * numOfObjects
)

Parameters:

result The result handle containing inference results.
numOfObjects [out] The number of detected objects.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultDetectedObjects`¶

(DET) Retrieves an array of detected objects from Object Detection.

int SaigeGetInferenceResultDetectedObjects (
    SaigeInferenceResult result,
    SaigeDetectedObject * detectedObjects
)

Parameters:

result The result handle containing inference results.
detectedObjects [out] Array of detected objects.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultNumOfSegmentedObjects`¶

Retrieves the number of segmented objects detected by Segmentation (SEG) or Anomaly Detection (IAD).

int SaigeGetInferenceResultNumOfSegmentedObjects (
    SaigeInferenceResult result,
    int * num
)

Parameters:

result The result handle containing inference results.
num [out] The number of detected segmented objects.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultSegmentedObjects`¶

Retrieves an array of segmented objects detected by Segmentation (SEG) or Anomaly Detection (IAD).

int SaigeGetInferenceResultSegmentedObjects (
    SaigeInferenceResult result,
    SaigeSegmentedObject * segObjects
)

Parameters:

result The result handle containing inference results.
segObjects [out] Array of detected segmented objects.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultMask`¶

Retrieves a Mask array from Segmentation (SEG) or Anomaly Detection (IAD).

int SaigeGetInferenceResultMask (
    SaigeInferenceResult result,
    unsigned char ** mask
)

(IAD) Pixels classified as normal are 0, while those classified as abnormal are 1.
(SEG) Pixels classified as OK class are 0, while those classified under other classes are represented by their class index.
The array shape is (image height H, image width W).
This is a single pointer 1D array with row-major order. For row i and column j, the value is data[W*i + j].

Parameters:

result The result handle containing inference results.
mask [out] The Mask array.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultScoremapSEG`¶

Retrieves the Scoremap array for Segmentation (SEG).

int SaigeGetInferenceResultScoremapSEG (
    SaigeInferenceResult result,
    unsigned char ** scoremap
)

Scoremap for SEG, which differs from that of IAD.
Shape is (number of classes including OK, image height H, image width W).
If pixel (i, j) is predicted as class c, then (c, i, j) contains the score as an integer in the range 0–255.
This is a single pointer 1D array with row-major order, where the value for class c, row i, column j is data[W*H*c + W*i + j].

Parameters:

result The result handle containing inference results.
scoremap [out] The Scoremap array.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultIsOversized`¶

Checks if the input image size exceeded the inspection size set in SaigeHandleConfig.

int SaigeGetInferenceResultIsOversized (
    SaigeInferenceResult result,
    bool * isOversized
)

Parameters:

result The result handle containing inference results.
isOversized [out] true if the input image size exceeded the inspection size, otherwise false.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultAnomalyScores`¶

Retrieves three types of scores for Anomaly Detection (IAD).

int SaigeGetInferenceResultAnomalyScores (
    SaigeInferenceResult result,
    double * anomalyScore,
    double * minScore,
    double * maxScore
)

Parameters:

result The result handle containing inference results.
anomalyScore [out] The representative anomaly score of the image.
minScore [out] The minimum pixel-level score in the image.
maxScore [out] The maximum pixel-level score in the image.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultIsNg`¶

Retrieves the normal/abnormal classification result for the image in Anomaly Detection (IAD).

int SaigeGetInferenceResultIsNg (
    SaigeInferenceResult result,
    bool * isNg
)

Parameters:

result The result handle containing inference results.
isNg [out] true if the image is classified as abnormal, false if normal.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultScoremapIAD`¶

Retrieves the Scoremap array for Anomaly Detection (IAD).

int SaigeGetInferenceResultScoremapIAD (
    SaigeInferenceResult result,
    float ** scoremap
)

Scoremap specific to IAD, different from SEG’s scoremap.
Each pixel's score is represented as a non-negative floating-point number.
The array shape is (image height H, image width W).
This is a single pointer 1D array with row-major order, where the value for row i and column j is data[W*i + j].

Parameters:

result The result handle containing inference results.
scoremap [out] The Scoremap array.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultHeatmap`¶

SaigeGetInferenceResultHeatmap.

int SaigeGetInferenceResultHeatmap (
    SaigeInferenceResult result,
    float ** heatmap
)

Returns a Heatmap array that is normalized between heatmapColorRangeMin and heatmapColorRangeMax. Each pixel value is represented as a floating-point number within the range 0 to 1.
The shape is (image height H, image width W).
Implemented as a single pointer 1D array in row-major order. Thus, the value at row i and column j is data[W*i + j].

Parameters:

result The result handle containing inference results.
heatmap [out] The Heatmap array.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeGetInferenceResultNumOfTextObjects`¶

SaigeGetInferenceResultNumOfTextObjects.

int SaigeGetInferenceResultNumOfTextObjects (
    SaigeInferenceResult result,
    int * num
)

Retrieves the number of detected text areas in the image analyzed by OCR.

Parameters:

result The handle containing inference results.
num [out] The number of detected text areas.

Returns:

errorCode Returns 0 on successful execution, or an error code if an error occurs.

`SaigeGetInferenceResultTextObjects`¶

SaigeGetInferenceResultTextObjects.

int SaigeGetInferenceResultTextObjects (
    SaigeInferenceResult result,
    SaigeTextObject * textObjects
)

Retrieves an array of detected text areas (SaigeTextObject) from OCR inference results.

Parameters:

result The handle containing inference results.
textObjects [out] The output array of SaigeTextObject structures. This array should be allocated with at least the number of elements returned by SaigeGetInferenceResultNumOfTextObjects.

Returns:

errorCode Returns 0 on successful execution, or an error code if an error occurs.

`SaigeGetInferenceResultEnhancedImage`¶

SaigeGetInferenceResultEnhancedImage.

int SaigeGetInferenceResultEnhancedImage (
    SaigeInferenceResult result,
    unsigned char ** image
)

Retrieves the enhanced quality image for Image Enhancement (IEN) results.

Parameters:

result The handle containing inference results.
image [out] The pointer to store the enhanced image data.

Returns:

errorCode Returns 0 on successful execution, or an error code if an error occurs.

`SaigeGetInferenceResultInspectionTime`¶

SaigeGetInferenceResultInspectionTime.

int SaigeGetInferenceResultInspectionTime (
    SaigeInferenceResult result,
    long long * imreadTime,
    long long * inferenceTime,
    long long * postProcessingTime
)

Retrieves the time spent on various stages of the inference API execution, measured in milliseconds.

Parameters:

result The handle containing inference results.
imreadTime [out] Time spent reading the image.
inferenceTime [out] Time spent on image preprocessing and deep learning inference.
postProcessingTime [out] Time spent on post-processing.

Returns:

errorCode Returns 0 on successful execution, or an error code if an error occurs.

`SaigeGetInferenceResultProductId`¶

SaigeGetInferenceResultProductId.

int SaigeGetInferenceResultProductId (
    SaigeInferenceResult result,
    const char ** productId,
    unsigned int * length
)

Retrieves the Product ID used or generated during inference execution.

Parameters:

result The handle containing inference results.
productId [out] Pointer to the Product ID string.
length [out] Length of the Product ID string.

Returns:

errorCode Returns 0 on successful execution, or an error code if an error occurs.

`SaigeInitializeAlignModule`¶

Function to activate the Saige Alignment function.

int SaigeInitializeAlignModule (
    const int gpuIndex,
    const int sourceWidth,
    const int sourceHeight,
    const int targetWidth,
    const int targetHeight
)

The Alignment function adjusts the position of the Source image to match the Target image and Crop it. Therefore, the Source image must be larger than the Target image.

Parameters:

gpuIndex This function uses the GPU, and is the index of the GPU to use.
sourceWidth
sourceHeight
targetWidth
targetHeight

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeFinalizeAlignModule`¶

Function to terminate the Saige Alignment function. Releases the GPU memory used by the function.

int SaigeFinalizeAlignModule ()

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeAlignImageFromByteArray`¶

Function to adjust the position of the Source image to match the Target image.

int SaigeAlignImageFromByteArray (
    const unsigned char * sourceImage,
    const int sourceWidth,
    const int sourceHeight,
    const int sourceChannel,
    const unsigned char * targetImage,
    const int targetWidht,
    const int targetHeight,
    const int targetChannel,
    unsigned int flag,
    unsigned char * croppedSourceImage
)

The croppedSourceImage should be targetWidth * targetHeight * sourceChannel

Parameters:

sourceImage The byte array of the Source image.
sourceWidth The width of the Source image.
sourceHeight The height of the Source image.
sourceChannel The channel of the Source image.
targetImage The byte array of the Target image.
targetWidht The width of the Target image.
targetHeight The height of the Target image.
targetChannel The channel of the Target image.
flag Whether to detect the edges of the cropped Source image.
croppedSourceImage The byte array of the cropped Source image.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

`SaigeAlignImageFromFile`¶

Not Implemented.

int SaigeAlignImageFromFile (
    const char * sourceImagePath,
    const char * targetImagePath,
    int * cropWidth,
    int * cropHeight,
    int * cropChannel,
    unsigned char * croppedSourceImage
)

`SaigeCreateInferenceHandle`¶

Creates an inference handle.

int SaigeCreateInferenceHandle (
    SaigeInferenceHandle * handle,
    const int projectType,
    const char * modelFilePath,
    const int deviceIndex,
    SaigeHandleConfig * config,
    const unsigned int flags
)

Ensure that the project type and model file path are correctly set according to the intended task. An error will occur if the specified project type and the project information in the model file do not match.

Parameters:

handle [out] The inference handle to use.
projectType Integer value representing the project type.
modelFilePath Path to the model file.
deviceIndex Index of the device to use for inference. Use the GPU index (non-negative integer) for GPU or -1 for CPU.
config Structure containing inference handle configuration settings.
flags Items to configure.

Returns:

errorCode Returns 0 on success, or an error code if an error occurs.

`SaigeDestroyInferenceHandle`¶

Destroys an inference handle.

int SaigeDestroyInferenceHandle (
    SaigeInferenceHandle handle
)

Parameters:

handle The inference handle that has completed usage.

Returns:

errorCode Returns 0 on success, or an error code if an error occurs.

`SaigeCreateInferenceResult`¶

Creates a result handle.

int SaigeCreateInferenceResult (
    SaigeInferenceResult * result
)

Parameters:

result [out] The result handle to use.

Returns:

errorCode Returns 0 on success, or an error code if an error occurs.

`SaigeDestroyInferenceResult`¶

Destroys a result handle.

int SaigeDestroyInferenceResult (
    SaigeInferenceResult result
)

Parameters:

result The result handle that has completed usage.

Returns:

errorCode Returns 0 on success, or an error code if an error occurs.

`SaigeGetModelInfo`¶

Retrieves model information from the inference handle.

int SaigeGetModelInfo (
    SaigeInferenceHandle handle,
    SaigeModelInfo * info
)

Parameters:

handle The inference handle from which to retrieve information.
info [out] Structure to store model information.

Returns:

errorCode Returns 0 on success, or an error code if an error occurs.

See also: SaigeModelInfo

`SaigeSetInferenceOption`¶

Sets inference option values on the inference handle.

int SaigeSetInferenceOption (
    SaigeInferenceHandle handle,
    SaigeInferenceOption * option,
    const unsigned int flags
)

Parameters:

handle The inference handle to use.
option The structure containing the option values to set.
flags The specific option items to set.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

See also: SaigeInferenceOption

`SaigeGetInferenceOption`¶

Retrieves inference option values from the inference handle.

int SaigeGetInferenceOption (
    SaigeInferenceHandle handle,
    SaigeInferenceOption * option
)

Parameters:

handle The inference handle from which to retrieve option values.
option [out] The structure to hold the retrieved option values.

Returns:

errorCode Returns 0 on success, or an error code if there is a failure.

See also: SaigeInferenceOption

`SaigeStrError`¶

SaigeStrError.

const char * SaigeStrError (
    const int errnum
)

Retrieves the message corresponding to a given SaigeVision error code.

Parameters:

errnum The error code.

Returns:

errorMessage Returns the error message associated with the error code.

`SaigeGetModuleVersion`¶

SaigeGetModuleVersion.

const char * SaigeGetModuleVersion ()

Retrieves the version of the modules used by SaigeVision.

Returns:

Format: (feature)version; (feature)version; ...

`SaigeSetHandleMetadata`¶

SaigeSetHandleMetadata.

int SaigeSetHandleMetadata (
    SaigeInferenceHandle handle,
    const char * key,
    const char * value
)

Sets metadata for the image collected during data collection. Once set, this metadata is retained.

Parameters:

handle The inference handle.
key The metadata key. Only keys recognized by the monitoring system will be reflected.
value The metadata value, interpreted according to the monitoring system's defined types.

`SaigeSetInferenceMetadata`¶

SaigeSetInferenceMetadata.

int SaigeSetInferenceMetadata (
    SaigeInferenceResult result,
    const char * key,
    const char * value
)

Sets one-time metadata for the image collected during data collection.

Parameters:

result The inference result handle.
key The metadata key. Only keys recognized by the monitoring system will be reflected.
value The metadata value, interpreted according to the monitoring system's defined types.

`SaigeSetResultMetadata`¶

SaigeSetResultMetadata.

int SaigeSetResultMetadata (
    SaigeInferenceResult result,
    const char * key,
    const char * value
)

Sets one-time test result metadata for the collected image during data collection.

Parameters:

result The inference result handle.
key The metadata key. Only keys recognized by the monitoring system will be reflected.
value The metadata value, interpreted according to the monitoring system's defined types.

Macro Definition Documentation¶

`SAIGE_WRITE_ON_MODEL_FILE`¶

(Common) When setting inference option values using SaigeSetInferenceOption, this flag saves the option values to the model file as well.

#define SAIGE_WRITE_ON_MODEL_FILE 0x01

`SAIGE_SET_CALC_TIME`¶

(Common) Flag to set the calcTime option.

#define SAIGE_SET_CALC_TIME 0x2

`SAIGE_SET_OVERSIZED_IMAGE_HANDLING`¶

(SEG, OCR, IEN) Flag to set the oversizedImageHandling option.

#define SAIGE_SET_OVERSIZED_IMAGE_HANDLING 0x4

`SAIGE_SET_ADDITIONAL_SCORES`¶

(CLS) Flag to set the additionalScores option.

#define SAIGE_SET_ADDITIONAL_SCORES 0x8

`SAIGE_SET_CALC_CLASS_ACTIVATION_MAP`¶

(CLS) Flag to set the calcClassActivationMap option.

#define SAIGE_SET_CALC_CLASS_ACTIVATION_MAP 0x10

`SAIGE_SET_MAX_NUM_OF_DETECTED_OBJECTS`¶

(DET) Flag to set the maxNumOfDetectedObjects option.

#define SAIGE_SET_MAX_NUM_OF_DETECTED_OBJECTS 0x20

`SAIGE_SET_CALC_SCOREMAP`¶

(SEG) Flag to set the calcScoremap option.

#define SAIGE_SET_CALC_SCOREMAP 0x40

`SAIGE_SET_CALC_OBJECT`¶

(SEG, IAD) Flag to set the calcObject option.

#define SAIGE_SET_CALC_OBJECT 0x80

`SAIGE_SET_CALC_OBJECT_AREA_AND_APPLY_THRESHOLD`¶

(SEG, IAD) Flag to set the calcObjectAreaAndApplyThreshold option.

#define SAIGE_SET_CALC_OBJECT_AREA_AND_APPLY_THRESHOLD 0x100

`SAIGE_SET_CALC_OBJECT_SCORE_AND_APPLY_THRESHOLD`¶

(SEG, IAD) Flag to set the calcObjectScoreAndApplyThreshold option.

#define SAIGE_SET_CALC_OBJECT_SCORE_AND_APPLY_THRESHOLD 0x200

`SAIGE_SET_OBJECT_BOUNDARY_THRESHOLD`¶

(SEG, IAD) Flag to set objectBoundaryThresholdPerClass in SEG and objectBoundaryThreshold in IAD.

#define SAIGE_SET_OBJECT_BOUNDARY_THRESHOLD 0x400

`SAIGE_SET_OBJECT_AREA_THRESHOLD`¶

(SEG, DET, IAD) Flag to set objectAreaThresholdPerClass in SEG and DET, and objectAreaThreshold in IAD.

#define SAIGE_SET_OBJECT_AREA_THRESHOLD 0x800

`SAIGE_SET_OBJECT_SCORE_THRESHOLD`¶

(SEG, DET, IAD) Flag to set objectScoreThresholdPerClass in SEG and DET, and objectScoreThreshold in IAD.

#define SAIGE_SET_OBJECT_SCORE_THRESHOLD 0x1000

`SAIGE_SET_CALC_MASK`¶

(SEG, IAD) Flag to set the calcMask option.

#define SAIGE_SET_CALC_MASK 0x2000

`SAIGE_SET_CALC_HEATMAP`¶

(IAD) Flag to set the calcHeatmap option.

#define SAIGE_SET_CALC_HEATMAP 0x4000

`SAIGE_SET_ANOMALY_SCORE_THRESHOLD`¶

(IAD) Flag to set the anomalyScoreThreshold option.

#define SAIGE_SET_ANOMALY_SCORE_THRESHOLD 0x8000

`SAIGE_SET_HEATMAP_COLOR_RANGE`¶

(IAD) Flag to set heatmapColorRangeMin and heatmapColorRangeMax options.

#define SAIGE_SET_HEATMAP_COLOR_RANGE 0x10000

`SAIGE_SET_POLYGON_SCORE_THRESHOLD`¶

(OCR) Flag to set the polygonScoreThreshold option.

#define SAIGE_SET_POLYGON_SCORE_THRESHOLD 0x20000

`SAIGE_SET_TEXT_SIZE_RANGE`¶

(OCR) Flag to set textSizeRangeMin and textSizeRangeMax options.

#define SAIGE_SET_TEXT_SIZE_RANGE 0x40000

`SAIGE_SET_NUM_OF_TEXT_OBJECTS`¶

(OCR) Flag to set the maxNumOfTextObjects option.

#define SAIGE_SET_NUM_OF_TEXT_OBJECTS 0x80000

`SAIGE_SET_TEXT_SCORE_THRESHOLD`¶

(OCR) Flag to set the textScoreThreshold option.

#define SAIGE_SET_TEXT_SCORE_THRESHOLD 0x100000

`SAIGE_SET_TEXT_FORMATTING`¶

(OCR) Flag to set the textFormatting option.

#define SAIGE_SET_TEXT_FORMATTING 0x200000

`SAIGE_IMAGE_COLOR`¶

Flag indicating color image data.

#define SAIGE_IMAGE_COLOR 0x01

`SAIGE_IMAGE_16BIT`¶

Flag indicating 16-bit per channel image data.

#define SAIGE_IMAGE_16BIT 0x02

`SAIGE_IMAGE_ALPHA`¶

Flag indicating data with an alpha channel.

#define SAIGE_IMAGE_ALPHA 0x04

`SAIGE_IMAGE_BGR`¶

Flag indicating the color channel order is Blue, Green, Red.

#define SAIGE_IMAGE_BGR 0x08

`SAIGE_IMAGE_8BIT_GRAY`¶

Macro indicating 8-bit grayscale image data. It is a single-channel array with a size of Width x Height.

#define SAIGE_IMAGE_8BIT_GRAY 0x00

`SAIGE_IMAGE_24BIT_RGB`¶

Macro indicating 24-bit RGB color image data. It is an array of size Width x Height x Channels (3) with channels ordered as R, G, B.

#define SAIGE_IMAGE_24BIT_RGB ( SAIGE_IMAGE_COLOR )

`SAIGE_IMAGE_24BIT_BGR`¶

Macro indicating 24-bit BGR color image data. It is an array of size Width x Height x Channels (3) with channels ordered as B, G, R.

#define SAIGE_IMAGE_24BIT_BGR ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_BGR )

`SAIGE_IMAGE_32BIT_RGBA`¶

Macro indicating 32-bit RGBA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as R, G, B, A.

#define SAIGE_IMAGE_32BIT_RGBA ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_ALPHA )

`SAIGE_IMAGE_32BIT_BGRA`¶

Macro indicating 32-bit BGRA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as B, G, R, A.

#define SAIGE_IMAGE_32BIT_BGRA ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_ALPHA | SAIGE_IMAGE_BGR )

`SAIGE_IMAGE_16BIT_GRAY`¶

Macro indicating 16-bit grayscale image data. It is a single-channel array of size Width x Height, with each pixel represented as an unsigned short.

#define SAIGE_IMAGE_16BIT_GRAY ( SAIGE_IMAGE_16BIT )

`SAIGE_IMAGE_48BIT_RGB`¶

Macro indicating 48-bit RGB color image data. It is an array of size Width x Height x Channels (3) with channels ordered as R, G, B, and each pixel represented as an unsigned short.

#define SAIGE_IMAGE_48BIT_RGB ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_16BIT )

`SAIGE_IMAGE_48BIT_BGR`¶

Macro indicating 48-bit BGR color image data. It is an array of size Width x Height x Channels (3) with channels ordered as B, G, R, and each pixel represented as an unsigned short.

#define SAIGE_IMAGE_48BIT_BGR ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_16BIT | SAIGE_IMAGE_BGR )

`SAIGE_IMAGE_64BIT_RGBA`¶

Macro indicating 64-bit RGBA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as R, G, B, A, and each pixel represented as an unsigned short.

#define SAIGE_IMAGE_64BIT_RGBA ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_16BIT | SAIGE_IMAGE_ALPHA )

`SAIGE_IMAGE_64BIT_BGRA`¶

Macro indicating 64-bit BGRA color image data. It is an array of size Width x Height x Channels (4) with channels ordered as B, G, R, A, and each pixel represented as an unsigned short.

#define SAIGE_IMAGE_64BIT_BGRA ( SAIGE_IMAGE_COLOR | SAIGE_IMAGE_16BIT | SAIGE_IMAGE_ALPHA | SAIGE_IMAGE_BGR )

`SAIGEVISION_RUNTIME_VERSION_MAJOR`¶

Major version of the SAIGE VISION Runtime C/C++ SDK.

#define SAIGEVISION_RUNTIME_VERSION_MAJOR 2

`SAIGEVISION_RUNTIME_VERSION_MINOR`¶

Minor version of the SAIGE VISION Runtime C/C++ SDK.

#define SAIGEVISION_RUNTIME_VERSION_MINOR 12

`SAIGEVISION_RUNTIME_VERSION_PATCH`¶

Patch version of the SAIGE VISION Runtime C/C++ SDK.

#define SAIGEVISION_RUNTIME_VERSION_PATCH 3

`SV_STR`¶

#define SV_STR (
    x
) #x

`SV_VER`¶

#define SV_VER (
    M,
    m,
    p
) SV_STR(M) "." SV_STR(m) "." SV_STR(p)

`SAIGEVISION_RUNTIME_VERSION`¶

The version of the SaigeVision Runtime C/C++ SDK (deprecated). Please use MAJOR, MINOR, and PATCH versions instead.

#define SAIGEVISION_RUNTIME_VERSION "2.12.3"

`SAIGE_SET_BATCH_SIZE`¶

(CLS, DET, SEG, IAD) Flag to set batch size for inference. Used when batch inference API is applied.

#define SAIGE_SET_BATCH_SIZE 0x01

`SAIGE_SET_INSPECTION_SIZE`¶

(SEG, OCR) Flag to set the width and height of images for inference.

#define SAIGE_SET_INSPECTION_SIZE 0x02

`SAIGE_OPTIMIZE_HANDLE`¶

(Common) Flag to apply optimization to the inference handle. Applying inference optimization reduces inference time but may slightly alter results.

#define SAIGE_OPTIMIZE_HANDLE 0x04

`SAIGE_DO_NOT_INSPECT`¶

(SEG) An option for oversizedImageHandling, where inference is not performed and an error is raised.

#define SAIGE_DO_NOT_INSPECT "do_not_inspect"

`SAIGE_DO_NOT_ENHANCE`¶

(IEN) An option for oversizedImageHandling, where inference is not performed and an error is raised.

#define SAIGE_DO_NOT_ENHANCE "do_not_enhance"

`SAIGE_RESIZE_TO_FIT`¶

(SEG) An option for oversizedImageHandling. When set, the image is resized to fit within the inspection size, maintaining its aspect ratio, and then inference is performed.

#define SAIGE_RESIZE_TO_FIT "resize_to_fit"

`SAIGE_CROP_INTO_TILES`¶

(SEG, IEN) An option for oversizedImageHandling. When set, the image is cropped into tiles based on the inspection size. Inference is performed on each tile, and results are combined.

#define SAIGE_CROP_INTO_TILES "crop_into_tiles"

SaigeVisionAPI.h¶

Classes¶

Public Types¶

Public Attributes¶

Public Functions¶

Macros¶

Public Types Documentation¶

SaigeInferenceHandle¶

SaigeInferenceResult¶

Public Attributes Documentation¶

kProjectTypeCLS¶

kProjectTypeDET¶

kProjectTypeSEG¶

kProjectTypeIAD¶

kProjectTypeOCR¶

kProjectTypeIEN¶

kMaxLenOfClassName¶

kMaxNumOfClasses¶

kMaxLenOfModelName¶

kMaxLenOfCameraName¶

kMaxLenOfTrainedVersion¶

kMaxLenOfToolName¶

kMaxLenOfTextFormatting¶

Public Functions Documentation¶

SaigeInferFromImageFile¶

SaigeInferFromByteArray¶

SaigeInferFromByteArrayWithProductId¶

SaigeInferFromMultipleByteArray¶

SaigeInferFromMultiPageByteArray¶

SaigeGetInferenceResultScore¶

SaigeGetInferenceResultClass¶

SaigeGetInferenceResultClassActivationMap¶

SaigeGetInferenceResultNumOfDetectedObjects¶

SaigeGetInferenceResultDetectedObjects¶

SaigeGetInferenceResultNumOfSegmentedObjects¶

SaigeGetInferenceResultSegmentedObjects¶

SaigeGetInferenceResultMask¶

SaigeGetInferenceResultScoremapSEG¶

SaigeGetInferenceResultIsOversized¶

SaigeGetInferenceResultAnomalyScores¶

SaigeGetInferenceResultIsNg¶

SaigeGetInferenceResultScoremapIAD¶

SaigeGetInferenceResultHeatmap¶

SaigeGetInferenceResultNumOfTextObjects¶

SaigeGetInferenceResultTextObjects¶

SaigeGetInferenceResultEnhancedImage¶

SaigeGetInferenceResultInspectionTime¶

SaigeGetInferenceResultProductId¶

SaigeInitializeAlignModule¶

SaigeFinalizeAlignModule¶

SaigeAlignImageFromByteArray¶

SaigeAlignImageFromFile¶

SaigeCreateInferenceHandle¶

SaigeDestroyInferenceHandle¶

SaigeCreateInferenceResult¶

SaigeDestroyInferenceResult¶

SaigeGetModelInfo¶

SaigeSetInferenceOption¶

SaigeGetInferenceOption¶

SaigeStrError¶

SaigeGetModuleVersion¶

SaigeSetHandleMetadata¶

SaigeSetInferenceMetadata¶

SaigeSetResultMetadata¶

Macro Definition Documentation¶

SAIGE_WRITE_ON_MODEL_FILE¶

SAIGE_SET_CALC_TIME¶

SAIGE_SET_OVERSIZED_IMAGE_HANDLING¶

SAIGE_SET_ADDITIONAL_SCORES¶

SAIGE_SET_CALC_CLASS_ACTIVATION_MAP¶

SAIGE_SET_MAX_NUM_OF_DETECTED_OBJECTS¶

SAIGE_SET_CALC_SCOREMAP¶

SAIGE_SET_CALC_OBJECT¶

SAIGE_SET_CALC_OBJECT_AREA_AND_APPLY_THRESHOLD¶

SAIGE_SET_CALC_OBJECT_SCORE_AND_APPLY_THRESHOLD¶

SAIGE_SET_OBJECT_BOUNDARY_THRESHOLD¶

SAIGE_SET_OBJECT_AREA_THRESHOLD¶

SAIGE_SET_OBJECT_SCORE_THRESHOLD¶

SAIGE_SET_CALC_MASK¶

SAIGE_SET_CALC_HEATMAP¶

`SaigeInferenceHandle`¶

`SaigeInferenceResult`¶

`kProjectTypeCLS`¶

`kProjectTypeDET`¶

`kProjectTypeSEG`¶

`kProjectTypeIAD`¶

`kProjectTypeOCR`¶

`kProjectTypeIEN`¶

`kMaxLenOfClassName`¶

`kMaxNumOfClasses`¶

`kMaxLenOfModelName`¶

`kMaxLenOfCameraName`¶

`kMaxLenOfTrainedVersion`¶

`kMaxLenOfToolName`¶

`kMaxLenOfTextFormatting`¶

`SaigeInferFromImageFile`¶

`SaigeInferFromByteArray`¶

`SaigeInferFromByteArrayWithProductId`¶

`SaigeInferFromMultipleByteArray`¶

`SaigeInferFromMultiPageByteArray`¶

`SaigeGetInferenceResultScore`¶

`SaigeGetInferenceResultClass`¶

`SaigeGetInferenceResultClassActivationMap`¶

`SaigeGetInferenceResultNumOfDetectedObjects`¶

`SaigeGetInferenceResultDetectedObjects`¶

`SaigeGetInferenceResultNumOfSegmentedObjects`¶

`SaigeGetInferenceResultSegmentedObjects`¶

`SaigeGetInferenceResultMask`¶

`SaigeGetInferenceResultScoremapSEG`¶

`SaigeGetInferenceResultIsOversized`¶

`SaigeGetInferenceResultAnomalyScores`¶

`SaigeGetInferenceResultIsNg`¶

`SaigeGetInferenceResultScoremapIAD`¶

`SaigeGetInferenceResultHeatmap`¶

`SaigeGetInferenceResultNumOfTextObjects`¶

`SaigeGetInferenceResultTextObjects`¶

`SaigeGetInferenceResultEnhancedImage`¶

`SaigeGetInferenceResultInspectionTime`¶

`SaigeGetInferenceResultProductId`¶

`SaigeInitializeAlignModule`¶

`SaigeFinalizeAlignModule`¶

`SaigeAlignImageFromByteArray`¶

`SaigeAlignImageFromFile`¶

`SaigeCreateInferenceHandle`¶

`SaigeDestroyInferenceHandle`¶

`SaigeCreateInferenceResult`¶

`SaigeDestroyInferenceResult`¶

`SaigeGetModelInfo`¶

`SaigeSetInferenceOption`¶

`SaigeGetInferenceOption`¶

`SaigeStrError`¶

`SaigeGetModuleVersion`¶

`SaigeSetHandleMetadata`¶

`SaigeSetInferenceMetadata`¶

`SaigeSetResultMetadata`¶

`SAIGE_WRITE_ON_MODEL_FILE`¶

`SAIGE_SET_CALC_TIME`¶

`SAIGE_SET_OVERSIZED_IMAGE_HANDLING`¶

`SAIGE_SET_ADDITIONAL_SCORES`¶

`SAIGE_SET_CALC_CLASS_ACTIVATION_MAP`¶

`SAIGE_SET_MAX_NUM_OF_DETECTED_OBJECTS`¶

`SAIGE_SET_CALC_SCOREMAP`¶

`SAIGE_SET_CALC_OBJECT`¶

`SAIGE_SET_CALC_OBJECT_AREA_AND_APPLY_THRESHOLD`¶

`SAIGE_SET_CALC_OBJECT_SCORE_AND_APPLY_THRESHOLD`¶

`SAIGE_SET_OBJECT_BOUNDARY_THRESHOLD`¶

`SAIGE_SET_OBJECT_AREA_THRESHOLD`¶

`SAIGE_SET_OBJECT_SCORE_THRESHOLD`¶

`SAIGE_SET_CALC_MASK`¶

`SAIGE_SET_CALC_HEATMAP`¶

`SAIGE_SET_ANOMALY_SCORE_THRESHOLD`¶

`SAIGE_SET_HEATMAP_COLOR_RANGE`¶

`SAIGE_SET_POLYGON_SCORE_THRESHOLD`¶

`SAIGE_SET_TEXT_SIZE_RANGE`¶

`SAIGE_SET_NUM_OF_TEXT_OBJECTS`¶

`SAIGE_SET_TEXT_SCORE_THRESHOLD`¶

`SAIGE_SET_TEXT_FORMATTING`¶

`SAIGE_IMAGE_COLOR`¶

`SAIGE_IMAGE_16BIT`¶

`SAIGE_IMAGE_ALPHA`¶