Module backends

Backend abstraction for Edge Impulse inference

This module provides a trait-based abstraction that allows switching between different inference backends:

• EIM Backend: Binary communication with Edge Impulse model processes via Unix sockets
• FFI Backend: Direct FFI calls to the Edge Impulse C++ SDK

The backend system is designed to be extensible, allowing new inference engines to be added by implementing the InferenceBackend trait.

Modules

eim

EIM backend implementation

Enums

BackendConfig

Configuration for different backend types

Traits

InferenceBackend

Trait for inference backends

Functions

create_backend

Factory function to create the appropriate backend