Below is a compiled list of resources that can be useful for immersive computing research, including datasets, toolboxes, optimization libraries, simulators, and testbeds.

Table of Contents

1. Immersive Datasets
2. Immersive Computing Toolbox
3. Optimization Toolbox
4. Simulator/Emulator
5. Testbeds
6. Network Traces

Immersive Datasets

2D Image/Video

• Kodak True Color Image Dataset: A standard test suite for image compression testing.

• Ultra Video Group (UVG) 4k Video Dataset: A collection of high-resolution (4K) video sequences suitable for compression and quality assessment research.

• USCMediaCommLab 2k Video Dataset: Video dataset featuring 2K resolution clips, often used for benchmarking video processing algorithms.

Multiview Image/Video

• WILDTRACK Seven-Camera HD Dataset: A 7-camera HD dataset for multi-view detection in real environments.

• Multi-camera Multiple People Tracking Dataset: Focused on multi-camera tracking of multiple people in challenging scenarios.

• Duke Multi-Target Multi-Camera Tracking Dataset: A large-scale multi-camera tracking dataset often used in person re-identification.

• Assembly101: A large-scale multi-view video dataset for understanding procedural activities.

Volumetric

• THUman2.1 Dataset: High-quality human scans captured by a dense DLSR rig.

• PointAvatar: The PointAvatar dataset provides preprocessed 3D head avatar data, including synchronized video frames and mesh information.

• NHA: The Neural Head Avatars dataset includes preprocessed training data, head tracking results, and optimized head avatars for two subjects.

NeRF/GS Static or Dynamic Scene

• Neural 3D Video Synthesis Dataset: A dataset from Facebook Research for neural 3D video synthesis approaches.

• Synthetic 360-degree scenes: Synthetic environments used for NeRF training and evaluation.

• Forward-facing scenes: Commonly used for light field and NeRF-based reconstructions.

• Unbounded 360-degree outdoor scenes: Outdoor 360-degree scenes for NeRF training in unbounded environments.

• NeRSemble Dataset: The NeRSemble Dataset is a large-scale multi-view video dataset of facial performances.

• ReRF: Three multi-view videos for kpop, box, and sing scenarios.

Immersive Computing Toolbox

• CompressAI: A library and tools for end-to-end compression research using machine learning.

• FFMPEG: A complete, cross-platform solution to record, convert and stream audio and video.

• BPG: BPG (Better Portable Graphics) is a new image format. Its purpose is to replace the JPEG image format when quality or file size is an issue.

• SVT-AV1 Encoder: The Scalable Video Technology for AV1 (SVT-AV1 Encoder) is an AV1-compliant software encoder library.

• Multiview High Efficiency Video Coding (MV-HEVC): Extensions of HEVC for encoding multi-view video content.

• DRACO 3D Data Compression: A library for compressing and decompressing 3D geometric meshes and point clouds.

• VVC Reference Software: The official reference software for the Versatile Video Coding standard.

• Video Multi-Method Assessment Fusion (VMAF): A perceptual video quality assessment algorithm developed by Netflix.

• Learned Perceptual Image Patch Similarity Metric (LPIPS): A metric for perceptual image patch similarity using deep features.

• nerfstudio: A simplified end-to-end framework for creating, training, and testing Neural Radiance Fields.

• gsplat: Tools and documentation for generating 3D scenes via splatting techniques.

Optimization Toolbox

• Bayesian Optimization: A Python library for Bayesian optimization of black-box functions.

• Multi-objective Bayesian Optimization (MOBOpt): A library for multi-objective Bayesian optimization in Python.

• pymoo: Multi-objective Optimization in Python: A framework that supports multi-objective optimization, genetic algorithms, and more.

• Gurobi: A state-of-the-art solver for linear, integer, and quadratic programming.

Simulator/Emulator

• Mahimahi: A tool for recording and replaying HTTP traffic over emulated network conditions.

• ns-3: A discrete-event network simulator used for internet systems research and education.

• tc: A Linux command for controlling network traffic and simulating latency, loss, and bottlenecks.

• cell-emulation-util: A script based on Linux TC netem to emulate the latency, loss, and bandwidth of a real-world cellular network.

Testbeds

• Emulab: A network testbed that allows researchers to run experiments on various environments.

• Cloudlab: A flexible infrastructure allowing cloud computing research on diverse hardware.

• Google Cloud: A public cloud platform providing a range of services for computing, storage, machine learning, etc.

• Amazon AWS: On-demand cloud computing platforms and APIs by Amazon.

• ILLIXR: Illinois Extended Reality testbed or ILLIXR is the first fully open-source Extended Reality (XR) system and testbed.

Network Traces

• Federal Communications Commission: Broadband performance measurement data from various U.S. ISPs.

• 3G/HSDPA Mobile Dataset: Commute Path Bandwidth Traces from 3G Networks.

• Mahimahi Cellular: These trace files represent the time-varying capacity of U.S. cellular networks as experienced by a mobile user.

• Lumos: A dataset about the throughput and delivery time of adaptive video streaming, which was collected in real-world mobile networks from December 2019 to May 2021.

• Puffer: Anonymized video sessions and telemetry data from real Puffer users.

• Interactive Video Streaming: Example traces for interactive video streaming.