PROJECT OUTCOMES

Software

As a Horizon Europe project co-financed by the European Comission, XReco was obliged to publish its activities, milestones and ongoing processes.

This page lists tools and services developed by the XReco consortium that was provided as open source software.

Table of Contents

vitrivr-engine (Multimedia Retrieval Engine)

The XReco NMR backend is powered by the vitrivr-engine1,2 – an open-source multi-modal multimedia retrieval system. vitrivr-engine is part of the vitrivr stack at, which is developed at UNIBAS and has been around for several years. It provides basic backend functionality for the two main workflows: asset ingest and asset search. This includes but is not limited to media decoding, segmentation, feature analysis & extraction and the orchestration of retrieval queries..

XRCapsules

XRCapsules is a web-based open source-application for creating XR experiences. Its goal is to serve as an easy way to generate Demonstrations, place assets in space, and program the execution of asset animations and access to triggers in a very simple way, to finally package it and build experiences using native Unity applications.

3D model descriptors extraction

This component does the 3D model descriptor extraction in the XReco platform.

It implements an analysis service to extract descriptors for 3D models. Once a request job has been created, the results can be polled via a job status query.

XReco NewsTagger LoRa adapters

LoRa adapters for meta-llama/Llama-3.1-8B-Instruct (Built with Llama) obtained through a finetuning process (using LoRA technique) aimed at making the model capable of attributing meaningful tags to a news article written in Italian, English, German or Slovenian.