xnohat/vector-advanced-ai

Vector with Advanced AI compatibilities

vector-advanced-ai

Vector with Advanced AI compatibilities

Setup

wire-pod

Anki Servers are down. You need to first setup a local server using wire-pod.

vector-sdk

Setup the sdk using my fork.

Object Detection (Raspberry Pi Compatible)

The project now uses Hugging Face Transformers with OWL-ViT for zero-shot object detection, optimized for Raspberry Pi CPU execution.

Install dependencies:

pip install -r requirements.txt

Installation Steps

1. Install system dependencies (Debian/Ubuntu):

sudo apt-get update
sudo apt-get install -y python3-pip python3-dev libjpeg-dev zlib1g-dev

2. Install Python packages:

pip install -r requirements.txt
pip install -e vector-python-sdk/

Run sample

python3 app.py

Performance Notes

• First run will download the OWL-ViT model (~500MB)
• Detection speed on Raspberry Pi 4: ~2-5 seconds per frame
• For faster inference, consider using smaller images or the google/owlvit-base-patch16 model