Skip to content

Latest commit

 

History

History
110 lines (98 loc) · 8.96 KB

Param.md

File metadata and controls

110 lines (98 loc) · 8.96 KB

Param

A single parameter set for one classification, for example face, describing the behaviour of the AI. Each classification has default parameters that are set if these are empty.

Properties

Name Type Description Notes
classification ClassificationEnum The classification of the object, that the element refers to. The following is a tree representation of the prototypes and their containing classifications. Default parameter values are defined per classification object. The default values of 'min', 'max' and 'drawMode' are witten after the name of the classification object in that order. Use the prototype names to set parameters to all of the containing classifications. ##### nudityCheck > A prototype containing classifications for basic human attributes and sexual contents. This prototype and the containing classifications are activated by default * face (1, 3, 0) * hand (0, -1, 0) * foot (0, -1, 0) * footwear (0, -1, 0) * chest (0, -1, 0) * breast (0, 0, 2) * vulva (0, 0, 2) * penis (0, 0, 2) * vagina (0, 0, 2) * buttocks (0, 0, 2) * anus (0, 0, 2) * oral (0, 0, 2) * penetration (0, 0, 2) * toy (0, -1, 0) * bondage (0, -1, 0) * gag (0, -1, 0) ##### ageEstimation > Contains classification objects for estimating the age of a human face. This prototype activates nudityCheck * child (*) (0, 0, 6) * adult (*) (0, -1, 0) * senior (*) (0, -1, 0) * pose (*) (0, 0, 0) - The age can not be estimated, due to a pose that hides facial features. ##### illegalSymbols > Contains various classification objects of symbols that are prohibited in Germany (e.g. Hakenkreuz, SS, etc). These are grouped under a single name * illegalSymbols (0, 0, 1) ##### textRecognition > Contains letters, numbers and some symbols (e.g. @, #, etc) as classification objects. These are grouped under a single name * textRecognition (0, 6, 6) ##### attributesCheck > Contains classification objects for various attributes of the human face. This prototype activates nudityCheck * female (0, -1, 0) * male (0, -1, 0) * hair (0, -1, 0) * hairless (0, -1, 0) * beard (0, -1, 0) * moustache (0, -1, 0) * headpiece (0, -1, 0) * glasses (0, -1, 0) * sunglasses (0, -1, 0) * mask (0, -1, 0) * slimSizedFace (*) (0, -1, 0) * realSizedFace (*) (0, -1, 0) * plusSizedFace (*) (0, -1, 0) ##### bodyAttributes > Contains classification objects for determining the body size. * slimSized (0, -1, 0) * realSized (0, -1, 0) * plusSized (0, -1, 0) ##### nippleCheck > Contains classifications for determining if the object recognized as breast has a nipple. This prototype activates nudityCheck * noNipple (0, -1, 0) * hasNipple (0, 0, 0) ##### unwantedSubstances > Contains classification objects for different substances. * beer (0, -1, 0) * beerBottle (0, -1, 0) * beerCan (0, -1, 0) * wine (0, -1, 0) * wineBottle (0, -1, 0) * cocktail (0, -1, 0) * alcohol (0, -1, 0) * cannabis (0, 0, 0) * cigarette (0, 0, 0) * cocaine (0, 0, 0) * heroine (0, 0, 0) * coffee (0, -1, 0) ##### violenceCheck > Contains classification objects for violent items or acts. * camouflage (0, -1, 0) * club (0, -1, 0) * knife (0, 0, 0) * sword (0, 0, 0) * pistol (0, 0, 0) * rifle (0, 0, 0) * cannon (0, 0, 0) * fire (0, -1, 0) ##### selfieCheck > Classification objects representing various pre-defined poses for the selfie task. * hand2mouth (0, 1, 0) * point2nose (0, 1, 0) * relaxed (0, 1, 0) * point2chin (0, 1, 0) * hand2cheek (0, 1, 0) * excellent (0, 1, 0) * nice (0, 1, 0) * thinking (0, 1, 0) * thumbUp (0, 1, 0) * victory (0, 1, 0) * lookout (0, 1, 0) * fingerUp (0, 1, 0) * middleFinger (0, 0, 0) Classification objects that are marked with (*) are sub-classifications of face. Both face and the marked classification are affected by the given parameter values. Please be aware that the default values can be subject to change. This is due to the difficulty of recognizing certain objects e.g. objects that are classified as toy.
prototype String The prototype that the classification object belongs to. Setting the prototype will specify the parameters for the classification belonging to the prototype. [optional]
min Integer The minimum amount of objects allowed on the image. Setting the value to -1 will cause the AI to ignore this rule. For moving images this value is applied to a single frame. [optional]
max Integer The maximum amount of objects allowed on the image. Setting the value to -1 will cause the AI to ignore this rule. For moving images this value is applied to a single frame. [optional]
severity Integer Set a value to define the severity of a broken rule of the given classification object. [optional]
drawMode Integer Set if and how a classification object should be drawn over. * 0 - will draw nothing, * 1 - will draw a frame with class name surrounding the object, * 2 - will draw a filled rectangle that will mask the object, * 3 - is a combination between 1 and 2 (frame/name + mask), * 6 - will blur the object and * 7 - is a combination between 1 and 6 (frame/name + blur). [optional]
grey Integer A grey scale color used in combination of drawMode '2' or '3'. '0' will represent black, while the maximum '255' will be white. [optional]
scale Float Scale of the bounds around the classification object. Specify a value to increase or decrease the size of the bounds. This is applied to the resulting media as well as the JSON coordinates. [optional]
ignore Boolean A shorthand to ignore the classification object. This is equal to setting min=0, max=-1 and drawMode=0. [optional]

Enum: ClassificationEnum

Name Value
NUDITY_CHECK "nudityCheck"
FACE "face"
HAND "hand"
FOOT "foot"
FOOTWEAR "footwear"
CHEST "chest"
BREAST "breast"
VULVA "vulva"
PENIS "penis"
VAGINA "vagina"
BUTTOCKS "buttocks"
ANUS "anus"
ORAL "oral"
PENETRATION "penetration"
TOY "toy"
BONDAGE "bondage"
GAG "gag"
AGE_ESTIMATION "ageEstimation"
CHILD "child"
ADULT "adult"
SENIOR "senior"
POSE "pose"
ILLEGAL_SYMBOLS "illegalSymbols"
TEXT_RECOGNITION "textRecognition"
ATTRIBUTES_CHECK "attributesCheck"
FEMALE "female"
MALE "male"
HAIR "hair"
HAIRLESS "hairless"
BEARD "beard"
MOUSTACHE "moustache"
HEADPIECE "headpiece"
GLASSES "glasses"
SUNGLASSES "sunglasses"
MASK "mask"
SLIM_SIZED_FACE "slimSizedFace"
REAL_SIZED_FACE "realSizedFace"
PLUS_SIZED_FACE "plusSizedFace"
BODY_ATTRIBUTES "bodyAttributes"
SLIM_SIZED "slimSized"
REAL_SIZED "realSized"
PLUS_SIZED "plusSized"
NIPPLE_CHECK "nippleCheck"
NO_NIPPLE "noNipple"
HAS_NIPPLE "hasNipple"
UNWANTED_SUBSTANCES "unwantedSubstances"
BEER "beer"
BEER_BOTTLE "beerBottle"
BEER_CAN "beerCan"
WINE "wine"
WINE_BOTTLE "wineBottle"
COCKTAIL "cocktail"
ALCOHOL "alcohol"
CANNABIS "cannabis"
CIGARETTE "cigarette"
COCAINE "cocaine"
HEROINE "heroine"
COFFEE "coffee"
VIOLENCE_CHECK "violenceCheck"
CAMOUFLAGE "camouflage"
CLUB "club"
KNIFE "knife"
SWORD "sword"
PISTOL "pistol"
RIFLE "rifle"
CANNON "cannon"
FIRE "fire"
SELFIE_CHECK "selfieCheck"
HAND2MOUTH "hand2mouth"
POINT2NOSE "point2nose"
RELAXED "relaxed"
POINT2CHIN "point2chin"
HAND2CHEEK "hand2cheek"
EXCELLENT "excellent"
NICE "nice"
THINKING "thinking"
THUMB_UP "thumbUp"
VICTORY "victory"
LOOKOUT "lookout"
FINGER_UP "fingerUp"
MIDDLE_FINGER "middleFinger"