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.
Unsupervised Deep Homography: A Fast and Robust Homography Estimation Model
| Date | Stars |
|---|---|
| 2026-07-24 | 290 |
| 2026-07-25 | 290 |
| 2026-07-28 | 290 |
| 2026-07-30 | 290 |
| 2026-08-06 | 290 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Unsupervised Deep Homography: A Fast and Robust Homography Estimation Model
[This paper](https://arxiv.org/abs/1709.03966) will be presented in International Conference on Robotics and Automation (ICRA) 2018 (Brisbane, Australia) and appear in proceedings of IEEE Robotics and Automation Letters.
We devise an unsupervised learning algorithm that trains a Deep Convolutional Neural Network to estimate planar homographies. We compare the proposed algorithm to traditional feature-based and direct methods, as well as a corresponding supervised learning algorithm. Our empirical results demonstrate that compared to traditional approaches, the unsupervised algorithm achieves faster inference speed, while maintaining comparable or better accuracy and robustness to illumination variation. In addition, on both a synthetic dataset and representative real-world aerial dataset, our unsupervised method has superior adaptability and performance compared to the supervised deep learning method.
<!--  -->
<p align="center">
<img width="400" height="300" src="teaser.JPG">
</p>
## Citation
If you use this code for research please cite:
@InProceedings{nguyen2017unsupervised,
title={Unsupervised Deep Homography: A Fast and Robust Homography Estimation Model},
author={Nguyen, Ty and Chen, Steven W and Shivakumar, Shreyas S and Taylor, Camillo J and Kumar, Vijay},
booktitle={RA-L},
pages={},
year={2018},
organization={IEEE}
month = " ",
year = "2018",
url = "https://arxiv.org/abs/1709.03966"
}
## Installation
Building and using requires the following libraries and programs\
cuda 8.0.61 (required for gpu support) \
python 2.7.12 \
tensorflow 1.2.1 (or higher)\
opencv 3.4.0 (can be installed using: pip install opencv-python )\
We built our system on ubuntu 16.04. Tensorflow (CPU) and Tensorflow (GPU) can both work well; they are installed in virtualenv. Other methods to install tensorflow have not been tested.
Install required python packages (pip is required)
```bash
source virtualenv_name/bin/activate
pip install -r requirements.txt
```
## Build instructions
### Clone repo
```bash
git clone https://github.com/tynguyen/unsupervisedDeepHomographyRAL2018.git
```
## Trained Models
### Model trained on Synthetic Data
Download at
```bash
https://drive.google.com/drive/folders/1Y9oNgbJTrAdkgf5-T1xONtU9n2ZqwDta?usp=sharing
```
Then, store the synthetic_models to folder models
### Model trained on Aerial Image Data
```bash
https://drive.google.com/drive/folders/16RI7R0EVayiXfYoP2Ahhl4yN2sWhG76Z?usp=sharing
```
Note: you need to format your image data in a correct size in order to make use of this trained model. Please refer to the next sections to get how to format the raw images
## Preparing training dataset (synthetic)
#### Download MS-COCO dataset http://cocodataset.org/#download
We use 2014/Train to generate training data and 2014/Testing to generate test set.
Store them into RAW_DATA_PATH and TEST_RAW_DATA_PATH which are repositories declared in generating synthetic data.
### Generate synthetic dataset
In the file code/utils/gen_synthetic_data.py, set important parameters as follows
```bash
RHO = 45 # The maximum value of pertubation. The higher it is, the larger displacement between
# two generated images is.
DATA_NUMBER = 100000 # number of pair of synthetic images in training dataset
TEST_DATA_NUMBER = 5000 # number of pair of synthetic images in test dataset
IM_PER_REAL = 2 # Generate 2 different synthetic images from one single real image
# Size of synthetic image
HEIGHT = 240
WIDTH = 320
# Size of crop
PATCH_SIZE = 128
# Directories to MS-COCO images
RAW_DATA_PATH = "/Earthbyte/tynguyen/rawdata/train/" # Real images used for generating synthetic data
TEST_RAW_DATA_PATH = "/Earthbyte/tynguyen/rawdata/test/" # Real images used for generating test syExcerpt of 14,562 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:3c45d17e162fa7f1, topic:tensorflow, topic:training
matched fp:3c45d17e162fa7f1, topic:robotics, readme:robotics