Skip to content

NSFW

NudeNet

NudeNet uses a fine-tuned yolov8 model to detect concepts related to SFW/NSFW distinctions in images.

Parameters

  • threshold (default: 0.5): The minimum confidence required to recognize a concept in the image.

Output properties

  • image.has_nudity: Whether the image contains nudity or not.
  • image.nudenet: The raw output of the NudeNet model, of type NudeNetDetections
  • image.nudenet.value.detections: A list of dicts like { "class": "<label>", "score": 0.8 } which contain the detected concepts and their confidence scores.
  • image.nudenet.value.has_nudity: Whether the image contains nudity or not.
  • image.nudenet.value.has_female: Whether the image contains a female subject.
  • image.nudenet.value.has_male: Whether the image contains a male subject.
  • image.nudenet.value.has(label): Whether the score for a specified label exceeds the threshold.

Supported Labels

FEMALE_GENITALIA_COVERED
FACE_FEMALE
BUTTOCKS_EXPOSED
FEMALE_BREAST_EXPOSED
FEMALE_GENITALIA_EXPOSED
MALE_BREAST_EXPOSED
ANUS_EXPOSED
FEET_EXPOSED
BELLY_COVERED
FEET_COVERED
ARMPITS_COVERED
ARMPITS_EXPOSED
FACE_MALE
BELLY_EXPOSED
MALE_GENITALIA_EXPOSED
ANUS_COVERED
FEMALE_BREAST_COVERED
BUTTOCKS_COVERED

Example

dataset >> NudeNet