Hi Rust community!

I'm learning Rust and built my first ML project with Rust - a facial recognition API server.
I'd really appreciate feedback on my code, especially around safety, idiomatic Rust patterns, and performance optimizations.

Project: Age/Gender/Emotion prediction server
PR: Age/Gender/Emotion recognition server with PyTorch + Rus
Tech: Actix-web + tch-rs (PyTorch bindings) + image processing

What it does:

• Loads a PyTorch TorchScript model (~36MB CNN)
• Serves predictions via REST API (~19ms/request on CPU)
• Handles multipart uploads, JSON responses, error handling

I know there are probably many improvements to make - that's exactly what I'm looking for!
Any feedback on code quality, safety issues, or Rust best practices would be hugely helpful.

Thanks in advance!

1 post - 1 participant

Read full topic