DOCSEARCHbyAlgolia
Submit Docs

QoEVAVE Datasets

website·Indexed 29 days ago

https://qoevave.github.io/qexe

A collection of audiovisual datasets including the Saliency Dataset, CGI Scenes Dataset, and 360 Scene Dataset. The resources provide scene descriptions, asset information, and downloads for various environments such as urban settings, natural landscapes, and CGI scenes, featuring Ambisonics (4OA) audio in ambiX format.

Tokens
28.8K
Snippets
54
Records
181
Agent score
75%

What's inside QoEVAVE

  1. Overview of the QoEVAVE 360 Scene Dataset

    The QoEVAVE Scenes Database consists of 12 real-life nature and urban scenes designed for evaluating uni-/multimodal perception, cognition, behavior, and Quality of Experience (QoE) in controlled virtual environments. Each sequence has a target duration of 60 seconds.

  2. Overview of the QoEVAVE Saliency Database

    The Saliency Database contains head-tracking data captured at 50 Hz and scene descriptions obtained via verbalization tasks. It includes cybersickness scores assessed through the simulator sickness questionnaire. Saliency data covers 360 video viewing via head-mounted displays across three audio scenarios: no audio, mono audio, and 4th-order Ambisonics audio.

  3. Overview of the EVOLVE-QoE project

    EVOLVE-QoE (Ecological Validity Evaluation of Interactive Virtual Environments: A QoE Framework for Audiovisual Scenes) is a research project running from 2024 to 2027. It aims to create schemata for characterizing Interactive Virtual Environments (IVEs) and IVE systems/media in terms of Ecological Validity (EV), specifically within the context of auditory cognition and Quality of Experience (QoE). The project builds upon the conceptual QoE framework, QExE Tool, and VR scenes developed during the previous QoEVAVE project.

  4. Overview of the QoEVAVE CGI Scene Database

    The CGI Scene Database provides three scenes for 6 degrees-of-freedom (6DoF) VR created with computer-generated imagery. Visuals are rendered using the Unity High-Definition Rendering Pipeline (HDRP), and audio is implemented via the Meta XR Audio SDK. The database includes bespoke audio geometry for high-fidelity rendering and is available as individual Unity packages or a complete Unity project.

  5. Overview of the QoEVAVE CGI Scene Database

    The QoEVAVE CGI Scene Database provides three high-quality audiovisual scenes designed for six degrees-of-freedom (6DoF) VR. Unlike 3DoF video, these scenes include interactive audiovisual objects, static and triggered audio sources, and fully modelled acoustic geometry for advanced audio rendering. The three available scenes are: The Cave, The Cinema, and The Mansion.

  6. Overview of the QoEVAVE Saliency Dataset

    The QoEVAVE Saliency Dataset contains head-tracking data from subjects exploring 14 360° videos via a head-mounted display (HMD). The dataset covers three audio conditions: no audio, mono audio, and 4th-order head-tracked Ambisonics. It includes head rotation data, post-viewing verbal scene descriptions, and Simulator Sickness Questionnaire (SSQ) results for cybersickness assessment.

  7. Overview of the QoEVAVE project

    QoEVAVE (Quality of Experience Evaluation of Interactive Virtual Environments with Audiovisual Scenes) was a research project running from 2021 to 2024. Its primary goal was to develop the first QoE framework for IVEs by integrating methodologies from the VR community and QoE research. The project focused on closing gaps in audio and video quality evaluation, examining how to infer quality from human behavior in IVEs, and creating a taxonomy of IVE-quality evaluation methods to account for immersion and presence.

  8. Overview of the Quality and Experience Evaluation (QExE) tool

    QExE is a framework designed to streamline subjective experiments in virtual reality, specifically focusing on audiovisual quality and quality of experience (QoE) in interactive VR environments. Built using Unity and MaxMSP, it automates experiment administration, including test item generation, stimulus presentation, participant interaction, and data logging, all driven by a single configuration file. It supports 360° video, CGI scenes, object-based audio, and Ambisonics, and integrates five quality evaluation methods and two validated questionnaires.

  9. Overview of the QoEVAVE 360 Scene Database

    The 360 Scene Database contains 12 different scene recordings (26 sequences total) with an average duration of 60 seconds. It supports video resolutions up to 8k and audio up to 4th-order Ambisonics in AmbiX format (ACN channel ordering & SN3D normalization). Spatial and temporal video information is provided for all sequences. Data is available as audio-only (.wav), video-only (.mkv), or muxed audio and video (.mp4).

  10. Understand River01 scene saliency audio scenarios

    Saliency data for the River01 scene was captured using three distinct audio scenarios to evaluate subject behavior:

    • No audio: Subjects are provided with no audio or headphones.
    • Mono audio: Mono audio is delivered via headphones, derived from the first channel of the ambisonics audio file.
    • 4OA audio: 4th-order Ambisonics is delivered via headphones with real-time head tracking (rotation).

  11. Understand audio scenarios for Badminton05 saliency data

    Saliency data for the Badminton05 scene is captured under three distinct audio scenarios:

    • No audio: Subjects are provided with no audio or headphones.
    • Mono audio: Mono audio is delivered via headphones, derived from the first channel of the ambisonics audio file.
    • 4OA audio: 4th-order Ambisonics is delivered via headphones with real-time head tracking (rotation).

  12. Understand Lake01 scene saliency audio scenarios

    Saliency data for the Lake01 scene was captured using three distinct audio scenarios to evaluate subject response:

    • No audio: Subjects were provided with no audio or headphones.
    • Mono audio: Mono audio was delivered via headphones, derived from the first channel of the ambisonics audio file.
    • 4OA audio: 4th-order Ambisonics was delivered via headphones with real-time head tracking (rotation).