HN Showcase
Back to browse
Boosted LightFace – A Hybrid DNN and GBM Model for Facial Recognition

Boosted LightFace – A Hybrid DNN and GBM Model for Facial Recognition

by serengil·Mar 3, 2026·2 points·0 comments
Visit ProjectView on HN

AI Analysis

○Pass

Facial recognition ensemble paper, not a shipped product or reproducible codebase.

Strengths
  • •99.1% LFW accuracy with deliberate no-retraining prevents benchmark gaming.
  • •Hybrid DNN+GBM ensemble approach is technically sound for production-grade robustness.
Weaknesses
  • •Posted as academic journal article, not a tool, library, or reproducible GitHub repo.
  • •No code, no binaries, no demo—requires manual implementation from the paper alone.
Category
AI/ML
Target Audience

Computer vision researchers, facial recognition practitioners, academic citations

Post Description

Hey, I'm the author of this work and the maintainer of DeepFace.

In this paper, we ensemble the top 5 DeepFace backbones and feed them into a GBM model instead of relying on a single embedding model.

On LFW, this surpasses FaceNet512’s 98.4% accuracy and the reported human-level accuracy of 97.5%, achieving 99.1%.

We deliberately avoid retraining on LFW to prevent benchmark overfitting. The base models are trained on large-scale datasets, and we only learn a boosting layer on top of their similarity scores.

Similar Projects