Which facial recognition engines are supported?

It primarily focuses on the integration of DoubleTake with CompreFace for high-accuracy local facial embedding and matching.

Can I use this for wake word detection?

Yes, it includes deployment patterns for OpenWakeWord to enable custom wake word detection for voice satellites like Willow.

Does this skill work with Home Assistant?

Yes, it includes detailed configurations for integrating the Wyoming voice pipeline and MQTT event data directly into Home Assistant's Assist Pipeline.

What hardware acceleration does this skill support?

The skill provides specific configurations for AMD GPUs using ROCm and OpenVINO, as well as NVIDIA GPUs using CUDA for Whisper and object detection.

Is the voice processing local or cloud-based?

The perception pipeline is designed for 100% local processing using Whisper for speech-to-text and Piper for text-to-speech, ensuring maximum privacy.

Perception Pipeline

Name: Perception Pipeline
Author: Lobbi-Docs

byLobbi-Docs

0•

Data Science & ML

Deploys and manages advanced smart home perception layers including NVR, facial recognition, and local voice processing pipelines.

The Perception Pipeline skill provides specialized knowledge for building sovereign, local-first smart home environments using the Ahling Command Center architecture. It offers domain-specific guidance for configuring Frigate NVR with AMD ROCm hardware acceleration, implementing facial recognition via DoubleTake and CompreFace, and orchestrating private voice pipelines using Whisper, Piper, and OpenWakeWord. This skill is ideal for developers seeking to automate complex video and audio event processing while maintaining data privacy and high-performance local execution.

Key Features

01Implements local facial recognition using DoubleTake and CompreFace

02Configures Frigate NVR with AMD GPU (ROCm) hardware acceleration

03Sets up Wyoming-compatible voice pipelines for STT and TTS

04Manages MQTT-based event structures for real-time automation

05Provides Docker Compose templates for perception-layer services

060 GitHub stars

Use Cases

01Integrating privacy-focused voice satellites with Home Assistant

02Automating home security through facial recognition and custom MQTT events

03Building a secure, local-only video surveillance system with object detection

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lobbi-docs/claude perception-pipeline

For use in Claude.ai and ChatGPT

Download Skill