Haipeng Li, Kunming Luo, Shuaicheng Liu,

This is the official implementation of our ICCV2021 paper GyroFlow.

Our presentation video: [Youtube][Bilibili].

2023-07: Try our Journal Extended Version GyroFlow+: Gyroscope-Guided Unsupervised Deep Homography and Optical Flow Learning

• Requirements please refer torequirements.txt.

We provide a toy demo to illustrate the process of converting gyroscope readings (i.e., angular velocity in row, pith, yaw) into an homography at gyro-video-stabilization, welcme to play with it

2021.11.15: We release the GOF_Train that contains 2000 samples.

2023.07.10: We release the GHOF_Train that contains 9900 samples.

The download link is GoogleDrive. Put the data into ./dataset/GHOF_Train, and the contents of directories are as follows:

./dataset/GOF_Train
├── sample_0
│   ├── img1.png
│   ├── img2.png
│   ├── gyro_homo.npy
├── sample_1
│   ├── img1.png
│   ├── img2.png
│   ├── gyro_homo.npy
.....................
├── sample_9900
│   ├── img1.png
│   ├── img2.png
│   ├── gyro_homo.npy

2023.07.10: We release the GHOF_Clean&Final that contains 5 categories, as the benchmark is changed, we thus update the metrics.

The pretrained model can be found at GoogleDrive. Move the model to ./experiments/demo_experiment/exp_2/test_model_best.pth.

For quantitative evaluation, including input frames and the corresponding gyro readings, a ground-truth optical flow is required for each pair.

The download link is GoogleDrive. Move the file to ./dataset/GHOF_Clean.npy.

The most difficult cases are collected in GOF-Final.

The download link is GoogleDrive. Move the file to ./dataset/GHOF_Final.npy.

To train the model, you can just run:

python train.py --model_dir experiments

Load the pretrained checkpoint and run:

python test.py --model_dir experiments/demo_experiment/exp_2 --restore_file experiments/demo_experiment/exp_2/test_model_best.pth

If you think this work is useful for your research, please kindly cite:

@InProceedings{Li_2021_ICCV,
    author    = {Li, Haipeng and Luo, Kunming and Liu, Shuaicheng},
    title     = {GyroFlow: Gyroscope-Guided Unsupervised Optical Flow Learning},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2021},
    pages     = {12869-12878}
}

In this project we use (parts of) the official implementations of the following works:

• ARFlow
• UpFlow
• RAFT
• DeepOIS

We thank the respective authors for open sourcing their methods.