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.
Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set (CVPRW 2019)
| Date | Stars |
|---|---|
| 2026-07-24 | 2348 |
| 2026-07-25 | 2348 |
| 2026-07-28 | 2348 |
| 2026-07-30 | 2348 |
| 2026-08-06 | 2348 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
## Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set ## <p align="center"> <img src="/images/example.gif"> </p> ### **_\*\*\*07/20/2021: A [PyTorch implementation](https://github.com/sicxu/Deep3DFaceRecon_pytorch) which has much better performance and is much easier to use is available now. This repo will not be maintained in future. \*\*\*_** This is a tensorflow implementation of the following paper: Y. Deng, J. Yang, S. Xu, D. Chen, Y. Jia, and X. Tong, [Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set](https://arxiv.org/abs/1903.08527), IEEE Computer Vision and Pattern Recognition Workshop (CVPRW) on Analysis and Modeling of Faces and Gestures (AMFG), 2019. (**_Best Paper Award!_**) The method enforces a hybrid-level weakly-supervised training for CNN-based 3D face reconstruction. It is fast, accurate, and robust to pose and occlussions. It achieves state-of-the-art performance on multiple datasets such as FaceWarehouse, MICC Florence and BU-3DFE. ## Features ### ● Accurate shapes The method reconstructs faces with high accuracy. Quantitative evaluations (shape errors in mm) on several benchmarks show its state-of-the-art performance: |Method|FaceWareHouse|Florence|BU3DFE| |:---:|:---:|:---:|:---:| |[Tewari et al. 17](https://arxiv.org/abs/1703.10580)</center>|2.19±0.54|-|-| |[Tewari et al. 18](https://arxiv.org/abs/1712.02859)|1.84±0.38|-|-| |[Genova et al. 18](https://arxiv.org/abs/1806.06098)|-|1.77±0.53|-| |[Sela et al. 17](https://arxiv.org/abs/1703.10131)|-|-|2.91±0.60| |[PRN 18](https://arxiv.org/abs/1803.07835)|-|-|1.86±0.47| |Ours|**1.81±0.50**|**1.67±0.50**|**1.40±0.31**| (Please refer to our paper for more details about these results) ### ● High fidelity textures The method produces high fidelity face textures meanwhile preserves identity information of input images. Scene illumination is also disentangled to generate a pure albedo. <p align="center"> <img src="/images/albedo.png"> </p> ### ● Robust The method can provide reasonable results under extreme conditions such as large pose and occlusions. <p align="center"> <img src="/images/extreme.png"> </p> ### ● Aligned with images Our method aligns reconstruction faces with input images. It provides face pose estimation and 68 facial landmarks which are useful for other tasks. We conduct an experiment on AFLW_2000 dataset (NME) to evaluate the performance, as shown in the table below: <p align="center"> <img src="/images/alignment.png"> </p> |Method|[0°,30°]|[30°,60°]|[60°,90°]|Overall| |:---:|:---:|:---:|:---:|:---:| |[3DDFA 16](https://arxiv.org/abs/1511.07212)</center>|3.78|4.54|7.93|5.42| |[3DDFA+SDM 16](https://arxiv.org/abs/1511.07212)|3.43|4.24|7.17|4.94| |[Bulat et al. 17](https://arxiv.org/abs/1703.00862)|**2.47**|**3.01**|**4.31**|**3.26**| |[PRN 18](https://arxiv.org/abs/1803.07835)|2.75|3.51|4.61|3.62| |Ours|2.56|3.11|4.45|3.37| ### ● Easy and Fast Faces are represented with Basel Face Model 2009, which is easy for further manipulations (e.g expression transfer). ResNet-50 is used as backbone network to achieve over 50 fps (on GTX 1080) for reconstructions. ## Getting Started ### Testing Requirements ### - Reconstructions can be done on both Windows and Linux. However, we suggest running on Linux because the rendering process is only supported on Linux. - Python 3.6 (numpy, scipy, pillow, argparse). - Tensorflow 1.12. - [Basel Face Model 2009 (BFM09)](https://faces.dmi.unibas.ch/bfm/main.php?nav=1-0&id=basel_face_model). - [Expression Basis (transferred from Facewarehouse by Guo et al.)](https://github.com/Juyong/3DFace). The original BFM09 model does not handle expression variations so extra expression basis are needed. - [tf mesh renderer](https://github.com/google/tf_mesh_renderer/tree/ba27ea1798f6ee8d03ddbc52f42ab4241f9328bb). We use the library to render reconstruction images. **Note that the rendering tool can only be used o
Excerpt of 15,253 characters
Read on GitHub
89
18
2
2
Alexander
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:d0b50628147ab969, topic:tensorflow