galthran-wq / face-recognition-service / 22098053886

74%

Build Type

push

github

Committed by web-flow

Commit Message

Add face detection, embedding, and analysis endpoints with provider a… (#1)

* Add face detection, embedding, and analysis endpoints with provider abstraction

Implement a pluggable face provider architecture (ABC + InsightFace backend)
with detect/embed/analyze endpoints and batch variants. Includes concurrency
control via semaphore, GPU support via Dockerfile.gpu, and full test coverage.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Fix Docker build: add g++ for insightface Cython compilation

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Make GPU Dockerfile configurable for different CUDA versions

Use build args CUDA_TAG and ONNXRT_VERSION so users can target their
host CUDA toolkit, e.g.:
  docker build --build-arg CUDA_TAG=11.8.0-cudnn8-runtime-ubuntu22.04 \
    --build-arg ONNXRT_VERSION=1.17.1

Default updated to CUDA 12.6 + onnxruntime-gpu 1.21.0.
Added deadsnakes PPA for Python 3.12 on Ubuntu 22.04 base images.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Default GPU Dockerfile to CUDA 12.4 + onnxruntime-gpu 1.24.1

Matches host CUDA 12.4 toolkit version.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Fix GPU Dockerfile: add DEBIAN_FRONTEND=noninteractive, use uv pip

- Prevents tzdata interactive prompt from hanging the build
- Use uv pip instead of .venv/bin/pip (uv venvs don't include pip)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Apply ruff formatting

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Add ruff format check reminder to CLAUDE.md

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Skip GPU tests in CI since runners lack GPUs

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Fix batch errors, gender mapping, and GPU fixture

* Update README for face recognition service

Rewrite from generic template to document face detection, embedding,
and analysis endpoints, configuration, GPU support, and provider architecture.

Co-Author... (continued)

Run Details

253 of 297 new or added lines in 10 files covered. (85.19%)

1 existing line in 1 file now uncovered.

345 of 394 relevant lines covered (87.56%)

0.88 hits per line