Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
News: the 10k dataset is ready for download.
| Date | Stars |
|---|---|
| 2026-07-24 | 650 |
| 2026-07-25 | 650 |
| 2026-07-28 | 650 |
| 2026-07-30 | 655 |
| 2026-08-11 | 658 |
| 2026-08-18 | 660 |
| 2026-08-20 | 661 |
| 2026-08-21 | 662 |
| 2026-08-24 | 662 |
| 2026-08-25 | 663 |
| 2026-08-26 | 664 |
| 2026-08-27 | 665 |
| 2026-08-31 | 666 |
| 2026-09-02 | 667 |
| 2026-09-04 | 668 |
| 2026-09-06 | 669 |
| 2026-09-11 | 670 |
| 2026-09-14 | 671 |
| 2026-09-18 | 673 |
| 2026-09-20 | 673 |
Today
— stars today
This week
+3 stars this week
This month
+11 stars this month
Momentum
0.0
growth rate 0.45%/day
<div align="center"> # DL3DV-10K Dataset **DL3DV-10K is a dataset of real-world videos with scene annotations and camera parameters.** <!-- This repo helps you get ready to download all the DL3DV-10K dataset. --> *Note: DL3DV-10K is a student-led project. Contact [Lu Ling](mailto:[email protected]) for any questions*. <img src="imgs/teaser.jpg" width="1000px"> --- <p align="center"> <a href="#dataset-download">Dataset Download</a> • <a href="https://dl3dv-10k.github.io/DL3DV-10K/">Website</a> • <a href="#nvs-benchmark-training-results">NVS Benchmark Training Results</a> • <a href="#data-preparation">Data Preparation</a> • <a href="#license">License</a> • <a href="#issues">Issues</a> • <a href="#bibtex">BibTex</a> </p> </div> ## News 🔥🔥🔥 * **We provide [DL3DV-Evaluation](https://huggingface.co/datasets/DL3DV/DL3DV-Evaluation) for test / evaluation. Note: Scenes in the DL3DV-Evaluation are not covered in the DL3DV-10K dataset** * The first 7K of **DL3DV-3DGS** is processed by [FCGS](https://github.com/YihangChen-ee/FCGS) and is now available at [DL3DV-GS-960P](https://huggingface.co/datasets/DL3DV/DL3DV-GS-960P)! * [stability.ai](https://stability.ai/news/introducing-stable-virtual-camera-multi-view-video-generation-with-3d-camera-control) employs DL3DV for **camera control** video generation. * [Cosmos](https://github.com/NVIDIA/Cosmos) employs DL3DV for **camera control** post-training in the World Foundation Model. * [DepthSplat](https://github.com/cvg/depthsplat) further builds on top of DL3DV. Take a look at their work and processed [dataset](https://github.com/cvg/depthsplat/blob/main/DATASETS.md)! * To help you create teaser image/video, we released all the drone videos to DL3DV huggingface: [drone](https://huggingface.co/datasets/DL3DV/DL3DV-Drone). * **We released all 10K videos/images/poses to huggingface!** Remember to update the [download.py](https://github.com/DL3DV-10K/Dataset/blob/main/scripts/download.py) script. ## Abstract We have witnessed significant progress in deep learning-based 3D vision, ranging from neural radiance field (NeRF) based 3D representation learning to applications in novel view synthesis (NVS). However, existing scene-level datasets for deep learning-based 3D vision, limited to either synthetic environments or a narrow selection of real-world scenes, are quite insufficient. This insufficiency not only hinders a comprehensive benchmark of existing methods but also caps what could be explored in deep learning-based 3D analysis. To address this critical gap, we present DL3DV-10K, a large-scale scene dataset, featuring 51.2 million frames from 10,510 videos captured from 65 types of point-of-interest (POI) locations, covering both bounded and unbounded scenes, with different levels of reflection, transparency, and lighting. We conducted a comprehensive benchmark of recent NVS methods on DL3DV-10K, which revealed valuable insights for future research in NVS. In addition, we have obtained encouraging results in a pilot study to learn generalizable NeRF from DL3DV-10K, which manifests the necessity of a large-scale scene-level dataset to forge a path toward a foundation model for learning 3D representation. ## Key Feature - 10,510 multi-view scenes covering 51.2 million frames at 4k resolution. - 140 videos as Novel view synthesis (NVS) benchmark. - All videos are annotated by scene environment (indoor vs. outdoor), levels of reflection, transparency, and lighting. - Released samples include colmap calculated camera pose. - Benchmark videos offer trained parameters from the SOTA NVS methods, including 3D Gaussian Splatting, ZipNeRF, Mip-NeRF 360, Instant-NGP, and Nerfacto. ## NVS Benchmark Training Results We report the performances of the main STOA methods (2023 Fall) on our large-scale NVS benchmark. Here are the quantitative results. Please refer to our paper for more details (e.g. more quantitative and qualitative results.) <div align="center"> <img src="
Excerpt of 15,774 characters
Read on GitHub
79
38
2
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:5626951deb660aea, topic:computer-vision, topic:3d-reconstruction, readme:gaussian splatting
matched fp:5626951deb660aea, topic:deep-learning, topic:pytorch
matched fp:5626951deb660aea, name:dataset, desc:dataset, readme:dataset