News for hackers. Stuff that matters.

serengil

Show HN: DeepFace is now running on PyTorch alongside TensorFlow

https://github.com/serengil/deepface

In early 2020, tensorflow was the dominant deep learning framework. So, deepface was built to run on top of tensorflow. Thereafter, pytorch becomes the dominant framework for deep learning. We got many requests from community to support pytorch.In the latest release (v0.0.101), DeepFace can now run on top of PyTorch alongside TensorFlow!Users has been installing deepface as `pip install...

( Read More... | no comments | on HN )

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

https://dergipark.org.tr/en/pub/gujs/article/1794891

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...

( Read More... | no comments | 2 points | on HN )

Show HN: DeepFace now supports DB-backed vector search for face recognition

https://sefiks.com/2026/01/01/introducing-brand-new-face-recognition-in-deepface/

DeepFace previously relied on filesystem-based datasets and pickle caches for face search, which made it stateful and hard to expose via APIs.We introduced register and search functions that store embeddings in a backend DB (Postgres, Mongo, Weaviate, Neo4j).This makes face recognition stateless, horizontally scalable, and suitable for API use.For Postgres and Mongo, FAISS is used for ANN...

( Read More... | no comments | on HN )
We use cosine or Euclidean distances for embedding models to make hard classifications. But this has a big limitation: no measure of confidence and no interpretability.

Instead, building a logistic regression model can turn distances into percentage based confidence scores. This also accounted for how a small decrease in distance affects the confidence score—similar to how a derivative measures sensitivity.

Show HN: Vector Similarity Computation with Partially Homomorphic Encryption

https://sefiks.com/2025/03/04/vector-similarity-search-with-partially-homomorphic-encryption-in-python/

Excited to share an update to LightPHE, a lightweight partially homomorphic encryption library for Python. This update enables the encryption and decryption of multi-dimensional vector embeddings and the computation of encrypted cosine similarity between an encrypted vector and a plain vector when using an additively homomorphic encryption.This relies on a two-tower architecture, where one...

( Read More... | no comments | on HN )

Show HN: LightPHE: Lightweight Partially Homomorphic Encryption for Python

https://github.com/serengil/LightPHE

Homomorphic encryption enables computations to be performed directly on encrypted data without requiring access to the private key. This allows data to be securely stored in the cloud while still being processed using the cloud’s computational power. Meanwhile, the cloud remains unaware of the actual data it is handling.Although fully homomorphic encryption has become available in recent years,...

( Read More... | no comments | 4 points | on HN )
DeepFace is a leading open-source library for facial recognition and facial attribute analysis, and the de facto standard in Python. It wraps multiple state-of-the-art models that have reached — and even surpassed — human-level accuracy in recognizing faces.

By the numbers (as of early 2025): 15,000+ stars on GitHub; ~4 million installations via pip; 800+ citations in academic papers

Whether you're building a cutting-edge AI project or simply exploring facial recognition, DeepFace makes advanced capabilities accessible with just a few lines of code.