# Robust Video Matting (RVM)
English | 中文
Official repository for the paper [Robust High-Resolution Video Matting with Temporal Guidance](https://peterl1n.github.io/RobustVideoMatting/). RVM is specifically designed for robust human video matting. Unlike existing neural models that process frames as independent images, RVM uses a recurrent neural network to process videos with temporal memory. RVM can perform matting in real-time on any videos without additional inputs. It achieves **4K 76FPS** and **HD 104FPS** on an Nvidia GTX 1080 Ti GPU. The project was developed at [ByteDance Inc.](https://www.bytedance.com/)
## News
* [Nov 03 2021] Fixed a bug in [train.py](https://github.com/PeterL1n/RobustVideoMatting/commit/48effc91576a9e0e7a8519f3da687c0d3522045f).
* [Sep 16 2021] Code is re-released under GPL-3.0 license.
* [Aug 25 2021] Source code and pretrained models are published.
* [Jul 27 2021] Paper is accepted by WACV 2022.
## Showreel
Watch the showreel video ([YouTube](https://youtu.be/Jvzltozpbpk), [Bilibili](https://www.bilibili.com/video/BV1Z3411B7g7/)) to see the model's performance.
All footage in the video are available in [Google Drive](https://drive.google.com/drive/folders/1VFnWwuu-YXDKG-N6vcjK_nL7YZMFapMU?usp=sharing).
## Demo
* [Webcam Demo](https://peterl1n.github.io/RobustVideoMatting/#/demo): Run the model live in your browser. Visualize recurrent states.
* [Colab Demo](https://colab.research.google.com/drive/10z-pNKRnVNsp0Lq9tH1J_XPZ7CBC_uHm?usp=sharing): Test our model on your own videos with free GPU.
## Download
We recommend MobileNetv3 models for most use cases. ResNet50 models are the larger variant with small performance improvements. Our model is available on various inference frameworks. See [inference documentation](documentation/inference.md) for more instructions.
Framework
Download
Notes
PyTorch
rvm_mobilenetv3.pth
rvm_resnet50.pth
Official weights for PyTorch.
Doc
TorchHub
Nothing to Download.
Easiest way to use our model in your PyTorch project.
Doc
TorchScript
rvm_mobilenetv3_fp32.torchscript
rvm_mobilenetv3_fp16.torchscript
rvm_resnet50_fp32.torchscript
rvm_resnet50_fp16.torchscript
If inference on mobile, consider export int8 quantized models yourself.
Doc
ONNX
rvm_mobilenetv3_fp32.onnx
rvm_mobilenetv3_fp16.onnx
rvm_resnet50_fp32.onnx
rvm_resnet50_fp16.onnx
Tested on ONNX Runtime with CPU and CUDA backends. Provided models use opset 12.
Doc,
Exporter.
TensorFlow
rvm_mobilenetv3_tf.zip
rvm_resnet50_tf.zip
TensorFlow 2 SavedModel.
Doc
TensorFlow.js
rvm_mobilenetv3_tfjs_int8.zip
Run the model on the web.
Demo,
Starter Code
CoreML
rvm_mobilenetv3_1280x720_s0.375_fp16.mlmodel
rvm_mobilenetv3_1280x720_s0.375_int8.mlmodel
rvm_mobilenetv3_1920x1080_s0.25_fp16.mlmodel
rvm_mobilenetv3_1920x1080_s0.25_int8.mlmodel
CoreML does not support dynamic resolution. Other resolutions can be exported yourself. Models require iOS 13+.
s denotes
downsample_ratio.
Doc,
Exporter
All models are available in [Google Drive](https://drive.google.com/drive/folders/1pBsG-SCTatv-95SnEuxmnvvlRx208VKj?usp=sharing) and [Baidu Pan](https://pan.baidu.com/s/1puPSxQqgBFOVpW4W7AolkA) (code: gym7).
## PyTorch Example
1. Install dependencies:
```sh
pip install -r requirements_inference.txt
```
2. Load the model:
```python
import torch
from model import MattingNetwork
model = MattingNetwork('mobilenetv3').eval().cuda() # or "resnet50"
model.load_state_dict(torch.load('rvm_mobilenetv3.pth'))
```
3. To convert videos, we provide a simple conversion API:
```
…
```
4. Or write your own inference code:
```
…
```
5. The models and converter API are also available through TorchHub.
```python
# Load the model.
model = torch.hub.load("PeterL1n/RobustVideoMatting", "mobilenetv3") # or "resnet50"
# Converter API.
convert_video = torch.hub.load("PeterL1n/RobustVideoMatting", "converter")
```
Please see [inference documentation](documentation/inference.md) for details on `downsample_ratio` hyperparameter, more converter arguments, and more advanced usage.
## Training and Evaluation
Please refer to the [training documentation](documentation/training.md) to train and evaluate your own model.
## Speed
Speed is measured with `inference_speed_test.py` for reference.
| GPU | dType | HD (1920x1080) | 4K (3840x2160) |
| -------------- | ----- | -------------- |----------------|
| RTX 3090 | FP16 | 172 FPS | 154 FPS |
| RTX 2060 Super | FP16 | 134 FPS | 108 FPS |
| GTX 1080 Ti | FP32 | 104 FPS | 74 FPS |
* Note 1: HD uses `downsample_ratio=0.25`, 4K uses `downsample_ratio=0.125`. All tests use batch size 1 and frame chunk 1.
* Note 2: GPUs before Turing architecture does not support FP16 inference, so GTX 1080 Ti uses FP32.
* Note 3: We only measure tensor throughput. The provided video conversion script in this repo is expected to be much slower, because it does not utilize hardware video encoding/decoding and does not have the tensor transfer done on parallel threads. If you are interested in implementing hardware video encoding/decoding in Python, please refer to [PyNvCodec](https://github.com/NVIDIA/VideoProcessingFramework).
## Project Members
* [Shanchuan Lin](https://www.linkedin.com/in/shanchuanlin/)
* [Linjie Yang](https://sites.google.com/site/linjieyang89/)
* [Imran Saleemi](https://www.linkedin.com/in/imran-saleemi/)
* [Soumyadip Sengupta](https://homes.cs.washington.edu/~soumya91/)
## Third-Party Projects
* [NCNN C++ Android](https://github.com/FeiGeChuanShu/ncnn_Android_RobustVideoMatting) ([@FeiGeChuanShu](https://github.com/FeiGeChuanShu))
* [lite.ai.toolkit](https://github.com/DefTruth/RobustVideoMatting.lite.ai.toolkit) ([@DefTruth](https://github.com/DefTruth))
* [Gradio Web Demo](https://huggingface.co/spaces/akhaliq/Robust-Video-Matting) ([@AK391](https://github.com/AK391))
* [Unity Engine demo with NatML](https://hub.natml.ai/@natsuite/robust-video-matting) ([@natsuite](https://github.com/natsuite))
* [MNN C++ Demo](https://github.com/DefTruth/lite.ai.toolkit/blob/main/lite/mnn/cv/mnn_rvm.cpp) ([@DefTruth](https://github.com/DefTruth))
* [TNN C++ Demo](https://github.com/DefTruth/lite.ai.toolkit/blob/main/lite/tnn/cv/tnn_rvm.cpp) ([@DefTruth](https://github.com/DefTruth))