Pigmentation Result Scheme

Haut.AI provides an AI algorithm that estimates pigmented spots for different facial areas and the whole face.

Description

Pigmentation Algorithm

algorithm_tech_name

pigmentation

area_name

The algorithm returns metrics and sub-metrics for the following facial areas:

• face

• nose_upper_triangle

• right_cheek_wide

• left_cheek_wide

• forehead

• chin_wide

main_metric

The main metric is an overall score indicating the pigmented spots for each area_name

"main_metric": {
                "value": 95,
                "widget_type": "bad_good_line",
                "name": "Pigmentation score",
                "tech_name": "pigmentation_score",
                "widget_meta": null
            },

• value the pigmentation score ranges from [0,100]The higher the value of this parameter, the fewer pigmented spots.

• The score can be classified into six classes based on value:

  • (90,100] - Excellent

  • (80,90] - Great

  • (70,80] - Good

  • (50,70] - Average

  • (30, 50] - Poor

  • [0,30] - Bad

• "widget_type": "bad_good_line" indicates that a higher value is better

• "name": "Pigmentation score"

• "tech_name": "pigmentation_score"

sub_metrics

The algorithm returns sub-metrics for each area_name

"sub_metrics": [
                {
                    "value": 12,
                    "widget_type": "numeric",
                    "name": "Number of pigmentation spots",
                    "tech_name": "number_of_regions",
                    "widget_meta": null
                },
                {
                    "value": 0,
                    "widget_type": "density",
                    "name": "Pigmentation density score",
                    "tech_name": "density_score",
                    "widget_meta": null
                },
                {
                    "value": 0,
                    "widget_type": "density",
                    "name": "Average intensity of pigmentation",
                    "tech_name": "average_intensity_score",
                    "widget_meta": null
                },
                {
                    "value": 9,
                    "widget_type": "density",
                    "name": "Average size of pigmentation spots",
                    "tech_name": "average_region_size",
                    "widget_meta": null
                },
                {
                    "value": 5,
                    "widget_type": "density",
                    "name": "Standard deviation of pigmentation spots size",
                    "tech_name": "region_size_std",
                    "widget_meta": null
                }
            ]

• Number of pigmented spots estimates the number of detected pigmented spots

  • value the detected number of spots

  • "widget_type": "numeric"

  • "name": "Number of pigmented spots"

  • "tech_name": "number_of_regions"

• Pigmentation density score estimates the density of spots

  • value the density of pigmented spots

  • "widget_type": "density"

  • "name": "Pigmentation density score"

  • "tech_name": "density_score"

• Average size of pigmented spots estimates the average size of pigmented spots in comparison with the skin area

  • value

  • "widget_type": "density"

  • "name": "Average size of pigmentation spots"

  • "tech_name": "average_region_size"

• Standard deviation of pigmented spots' size estimates the standard deviation of pigmented spots' size

  • value the density of pigmented spots

  • "widget_type": "density"

  • "name": "Standard deviation of pigmentation spots size"

  • "tech_name": "region_size_std"

• Average intensity of pigmentation estimates the intensity of pigmented spots

  • value the higher the value of this parameter, the more severe the spots

  • "widget_type": "density"

  • "name": "Average intensity of pigmentation"

  • "tech_name": "average_intensity_score"

masks_restored

The algorithm returns a vectorized masks of pigmented spots for a face-aligned image

• "mask_type": "polygon_mask"

• features

  • geometry

    • "type": "MultiPoint"

  See more on Masks here

masks_original

The algorithm returns a vectorized mask of pigmented spots for an original image

• "mask_type": "polygon_mask"

• features

  • geometry

    • "type": "MultiPoint"

  See more on Masks here

Example (JSON)

25KB

pigmentation.json

pigmentation.json