Documentation
¶
Overview ¶
Package dupedetection provides functions to compute dupe detection fingerprints for specific image
Index ¶
- Constants
- Variables
- func Blomqvist(data1, data2 []float64) (float64, error)
- func ComputeRandomizedDependence(x, y []float64) float64
- func FromFloat32To64(input []float32) []float64
- func GetMemoizer() *memoize.Memoizer
- func HSIC(data1, data2 []float64) (float64, error)
- func HoeffdingD(data1, data2 []float64) (float64, error)
- func Kendall(data1, data2 []float64) (float64, error)
- func MI(data1, data2 []float64) (float64, error)
- func MeasureImageSimilarity(candidateImageFingerprint []float32, ...) (int, error)
- func Pearson(data1, data2 []float64) (float64, error)
- func Spearman(data1, data2 []float64) (float64, error)
- type ComputeConfig
- type Config
- type MemoizationImageData
Constants ¶
const ( // DefaultInputDir that dd-service monitors for the new file to generate fingerprints DefaultInputDir = "input" // DefaultOutputDir that dd-service uses to put new fingerprints DefaultOutputDir = "output" // DefaultDataFile is path to the SQLite database used by dd-service DefaultDataFile = "dupe_detection_image_fingerprint_database.sqlite" )
Variables ¶
var InterfaceTypeError = errors.Errorf("Calculation function returned value of unexpected type.")
InterfaceTypeError indicates unexpected variable type is returned by invoked correlation calculation func
Functions ¶
func ComputeRandomizedDependence ¶
ComputeRandomizedDependence computes RDC correlation between input arrays of data
func FromFloat32To64 ¶
FromFloat32To64 accepts input array of float32 values and returns float64 array of the input values
func GetMemoizer ¶
GetMemoizer returns memoizer used to cache correlations calculations results
func HoeffdingD ¶
HoeffdingD calculates HoeffdingD correlation between arrays of input data
func MeasureImageSimilarity ¶
func MeasureImageSimilarity(candidateImageFingerprint []float32, fingerprintsArrayToCompareWith [][]float64, memoizationData MemoizationImageData, config ComputeConfig) (int, error)
MeasureImageSimilarity calculates similarity between candidateImageFingerprint and each value in fingerprintsArrayToCompareWith
Types ¶
type ComputeConfig ¶
type ComputeConfig struct {
CorrelationMethodNameArray []string
StableOrderOfCorrelationMethods []string
UnstableOrderOfCorrelationMethods []string
CorrelationMethodsOrder string
PearsonDupeThreshold float64
SpearmanDupeThreshold float64
KendallDupeThreshold float64
RandomizedDependenceDupeThreshold float64
BlomqvistDupeThreshold float64
HoeffdingDupeThreshold float64
HoeffdingRound1DupeThreshold float64
HoeffdingRound2DupeThreshold float64
MIThreshold float64
RootDir string
NumberOfImagesToValidate int
TrimByPercentile bool
FileStorage storage.FileStorage
}
ComputeConfig contains configurable parameters to calculate AUPRC of image similariy measurement
func NewComputeConfig ¶
func NewComputeConfig() ComputeConfig
NewComputeConfig retirieves new ComputeConfig with default values
type Config ¶
type Config struct {
// input directory for monitoring the new file to generate fingerprints
InputDir string `mapstructure:"input_dir" json:"input_dir,omitempty"`
// output directory for using to put new fingerprints
OutputDir string `mapstructure:"output_dir" json:"output_dir,omitempty"`
// data directory for dupe detection
DataFile string `mapstructure:"data_file" json:"data_file,omitempty"`
}
Config contains settings of the dupe detection service
func (*Config) SetWorkDir ¶
SetWorkDir applies `workDir` to DataFile, OutputDir and InputDir if it was not specified as an absolute path.
type MemoizationImageData ¶
type MemoizationImageData struct {
SHA256HashOfFetchedImages []string
SHA256HashOfCurrentImage string
}
MemoizationImageData provides images data for memoization of correlation calculation methods results
Source Files