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.
LiteFlowNet: A Lightweight Convolutional Neural Network for Optical Flow Estimation, CVPR 2018 (Spotlight paper, 6.6%)
| Date | Stars |
|---|---|
| 2026-07-24 | 633 |
| 2026-07-25 | 633 |
| 2026-07-28 | 632 |
| 2026-07-30 | 632 |
| 2026-07-31 | 632 |
| 2026-08-06 | 632 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# LiteFlowNet <p align="center"><img src="./figure/LiteFlowNet.png" width="800" /></p> <p align = "center">The network structure of LiteFlowNet. For the ease of representation, only a 3-level design is shown.</p> <p align="center"><img src="./figure/cascaded_flow_inference.png" width="400" /></p> <p align = "center">A cascaded flow inference module M:S in NetE.</p> This repository (<strong>https://github.com/twhui/LiteFlowNet</strong>) is the offical release of <strong>LiteFlowNet</strong> for my paper <a href="https://arxiv.org/pdf/1805.07036.pdf"><strong>LiteFlowNet: A Lightweight Convolutional Neural Network for Optical Flow Estimation</strong></a> in CVPR 2018 (Spotlight paper, 6.6%). <i>The up-to-date version of the paper is available on <a href="https://arxiv.org/pdf/1805.07036.pdf"><strong>arXiv</strong></a></i>. LiteFlowNet is a lightweight, fast, and accurate opitcal flow CNN. We develop several specialized modules including (1) pyramidal features, (2) cascaded flow inference (cost volume + sub-pixel refinement), (3) feature warping (f-warp) layer, and (4) flow regularization by feature-driven local convolution (f-lconv) layer. LiteFlowNet outperforms PWC-Net (CVPR 2018) on KITTI and has a smaller model size (less than PWC-Net by ~40%). For more details about LiteFlowNet, you may visit <a href="http://mmlab.ie.cuhk.edu.hk/projects/LiteFlowNet/"><strong>my project page</strong></a>. <a href="https://www.youtube.com/watch?v=pfQ0zFwv-hM"> <p align="center"><img src="./figure/demo video thumbnail.png" width="600" /></p> </a> Oral presentation at CVPR 2018 is also available on <a href="http://www.youtube.com/watch?v=LBJ20kxr1a0&t=60m33s"> <strong>YouTube</strong></a>. </ul> <table> <thead> <tr> <th align="center"></th> <th align="center">KITTI12 Testing Set (Out-Noc)</th> <th align="center">KITTI15 Testing Set (Fl-all)</th> <th align="center">Model Size (M)</th> </tr> <tr> <td align="center">FlowNet2 (CVPR17)</td> <td align="center">4.82%</td> <td align="center">10.41%</td> <td align="center">162.49</td> <tr> <td align="center">PWC-Net (CVPR18)</td> <td align="center">4.22%</td> <td align="center">9.60%</td> <td align="center">8.75</td> </tr> <tr> <td align="center"><strong>LiteFlowNet (CVPR18)</strong></td> <td align="center"><strong>3.27%</strong></td> <td align="center"><strong>9.38%</strong></td> <td align="center"><strong>5.37</strong></td> </tr> </tbody></table> # LiteFlowNet2 <strong>NEW! Our extended work (LiteFlowNet2, TPAMI 2020) is now available at https://github.com/twhui/LiteFlowNet2</strong>. LiteFlowNet2 in TPAMI 2020, another lightweight convolutional network, is evolved from LiteFlowNet (CVPR 2018) to better address the problem of optical flow estimation by improving flow accuracy and computation time. Comparing to our earlier work, LiteFlowNet2 improves the optical flow accuracy on Sintel clean pass by 23.3%, Sintel final pass by 12.8%, KITTI 2012 by 19.6%, and KITTI 2015 by 18.8%. Its runtime is 2.2 times faster! </ul> <table> <thead> <tr> <th align="center"></th> <th align="center">Sintel Clean Testing Set</th> <th align="center">Sintel Final Testing Set</th> <th align="center">KITTI12 Testing Set (Out-Noc)</th> <th align="center">KITTI15 Testing Set (Fl-all)</th> <th align="center">Model Size (M)</th> <th align="center">Runtime* (ms) GTX 1080</th> </tr> <tr> <td align="center">FlowNet2 (CVPR17)</td> <td align="center">4.16</td> <td align="center">5.74</td> <td align="center">4.82%</td> <td align="center">10.41%</td> <td align="center">162</td> <td align="center">121</td> </tr> <tr> <td align="center">PWC-Net+</td> <td align="center"><strong>3.45</strong></td> <td align="center"><strong>4.60</strong></td> <td align="center">3.36%</td> <td align="center">7.72% <td align="center">8.75</td> <td align="center"><strong>40</strong></td> </tr> <tr> <td align="center"><strong>LiteFlowNet2</strong></td> <td align="center"><strong>3.48</strong></td> <td align="center"><strong>4.69</strong></
Excerpt of 14,910 characters
Read on GitHub
179
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:4faed07f7edd54df, topic:pytorch, topic:tensorflow
matched fp:4faed07f7edd54df, topic:computer-vision, desc:optical flow, readme:optical flow