Ph-Eye

Ph-Eye is a service for hosting AI/NLP models for the purposes of finding PII and PHI in text.

Ph-Eye provides a simple REST API that accepts text and returns a list of identified entities. Each Docker image bundles a single model selected at build time. Images are designed for air-gapped deployment — the model is baked into the image during docker build and no network access is required at runtime.

Current Version: 1.2.5

While Ph-Eye can be used standalone, it was designed and created for use with Phileas and Philter and as such provides tight integration with each.

Available models

`PHEYE_MODEL`	Language	Entities	Underlying model
`pii_base`	English	General PII (person, place, org, …)	`philterd/ph-eye-pii-base`
`hospitals`	English	Hospital, room number	`knowledgator/gliner-pii-base-v1.0`
`medical_conditions`	English	Disease/disorder	`blaze999/Medical-NER`
`french_persons`	French	Person	`EmergentMethods/gliner_medium_news-v2.1`
`french_medical`	French	Disease (`Maladie`)	`almanach/camembert-bio-gliner-v0.1`

Features

Multiple models: Choose the model best suited to your use case at build time.
Air-gapped deployment: Models are downloaded at image build time — no outbound network access needed at runtime.
GPU support: GPU-enabled images are available via Dockerfile.gpu for accelerated inference with CUDA 12.1.
REST API: Simple JSON-based API for easy integration.
Dockerized: Easily deployable as a Docker container.
Integration: Works seamlessly with Phileas and Philter.

License

Ph-Eye is licensed under the Apache License, version 2.0. The model bundled in each image may be licensed separately - refer to the model's Hugging Face page for details.