English – The Conversation | Powerful AI is making facial recognition better at identifying you by Vijayan Asari, Professor of Electrical and Computer Engineering, University of Dayton
AI generated summary, Read the full article for complete information.
Facial‑recognition technology is becoming markedly more accurate thanks to powerful AI models trained on hundreds of millions of images, now achieving over 99 % correctness in controlled settings such as smartphones, airports and border checkpoints. Recent advances—like Google’s FaceNet, Facebook’s DeepFace, and NEC’s NeoFace—allow systems to recognize partially obscured faces, handle variations in lighting and expression, and even capture subtle facial dynamics across video frames, reducing both false‑positive and false‑negative errors. However, real‑world performance can still suffer from poor image quality, masks, makeup, and demographic imbalances in training data, which have historically led to higher misidentification rates for women, people of color, children and the elderly. Researchers are addressing these biases by diversifying datasets, applying image‑enhancement preprocessing, and developing three‑dimensional and volumetric directional‑pattern techniques that better mimic human facial perception, promising fewer wrongful arrests and a fairer, more reliable deployment of facial‑recognition systems.
Read more:
https://theconversation.com/powerful-ai-is-making-facial-recognition-better-at-identifying-you-276743
#Google #Meta #NEC
