AI Can Detect and Grade Acne Severity from Phone Photos

Acne is one of the most common skin concerns worldwide, and one of the hardest to assess consistently. Two dermatologists looking at the same face can disagree on severity grade. A 2022 study published in Diagnostics built a system called AcneDet that brings objectivity to the process.

Using Faster R-CNN for lesion detection combined with LightGBM for severity grading, AcneDet achieved 85% mean accuracy for acne severity classification on the IGA scale from smartphone photos. That is 18 percentage points higher than the previous benchmark of 67%.

The system does not just estimate overall severity. It detects and classifies individual lesion types - blackheads, whiteheads, acne scars, papules, pustules, and nodules - giving a granular picture of what is happening on the skin. This level of detail is exactly what estheticians need to recommend targeted treatments.

For salons and skincare professionals, this means AI can provide an objective baseline assessment of acne that clients can track over time. Before and after data becomes quantifiable, not just visual. Treatment recommendations become evidence-based, not opinion-based.