FaceNet
The 2015 triplet-loss embedding that made modern face recognition possible: 128 bytes per face, 99.63% on LFW.
About
FaceNet was Google's 2015 CVPR system for learning a direct mapping from face images into a compact Euclidean space where distances measure face similarity. Instead of a bottleneck classifier, it trained a deep CNN with online triplet mining, so verification, recognition, and clustering could all run on top of one embedding with standard techniques like k-NN.
The original Google weights were never released, but FaceNet survives through faithful community reimplementations (notably davidsandberg/facenet, MIT-licensed) trained on public data such as CASIA-WebFace and VGGFace2. It remains the reference architecture that nearly every later margin-loss model cites, even if the checkpoints themselves are now outpaced by ArcFace-family models.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported: 99.63% on LFW and 95.12% on YouTube Faces DB (paper, 2015), cutting error rate by ~30% vs the prior best. The original Google-trained weights were never released, so no independent replication of those exact numbers exists; community checkpoints report lower numbers on the same datasets.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified, metrics-v1). Evidence: peer-reviewed CVPR 2015 paper (arXiv 1503.03832) by named Google researchers; the triplet-loss formulation is foundational and cited thousands of times; community weights and MIT code are genuinely open. Strengths: rock-solid provenance, open weights and code, enormous adoption. Docked for: (1) the original trained weights were never released — only community retrains are available; (2) superseded in accuracy by newer margin-loss models. License: MIT (community code); paper is green open access.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 10.0/10
- Maintenance · recency, releases, issue load
- 5.0/10
- Openness · license + docs
- 8.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · CVPR 2015 paper (arXiv 1503.03832), 99.63% LFW provider-reported.
- · Foundational triplet-loss embedding; cited by nearly all later margin-loss work.
- · Only community retrains available; superseded by ArcFace-family models.