Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/sensorsini/ddd20-utils
DDD20 End-to-End Event Camera Driving Dataset
https://github.com/sensorsini/ddd20-utils
davis deep-learning event-camera neuromorphic-engineering python27
Last synced: 4 days ago
JSON representation
DDD20 End-to-End Event Camera Driving Dataset
- Host: GitHub
- URL: https://github.com/sensorsini/ddd20-utils
- Owner: SensorsINI
- License: lgpl-3.0
- Created: 2020-02-22T22:00:12.000Z (almost 5 years ago)
- Default Branch: master
- Last Pushed: 2024-11-15T17:48:33.000Z (about 1 month ago)
- Last Synced: 2024-12-20T23:22:54.991Z (5 days ago)
- Topics: davis, deep-learning, event-camera, neuromorphic-engineering, python27
- Language: Jupyter Notebook
- Homepage: https://sites.google.com/view/davis-driving-dataset-2020/home
- Size: 10.8 MB
- Stars: 23
- Watchers: 5
- Forks: 11
- Open Issues: 6
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# DDD20 End-to-End Event Camera Driving Dataset
See https://sites.google.com/view/davis-driving-dataset-2020/home for details.
Software released as part of the publication
* Hu, Y., Binas, J., Neil, D., Liu, S.-C., and Delbruck, T. (2020). "DDD20 End-to-End Event Camera Driving Dataset: Fusing Frames and Events with Deep Learning for Improved Steering Prediction". Special session Beyond Traditional Sensing for Intelligent Transportation, The 23rd IEEE International Conference on Intelligent Transportation Systems, September 20 – 23, 2020, Rhodes, Greece. arXiv [cs.CV]. arXiv. http://arxiv.org/abs/2005.08605
* Binas, J., Neil, D., Liu, S.-C., and Delbruck, T. (2017). DDD17: **End-To-End DAVIS Driving Dataset**. in _ICML’17 Workshop on Machine Learning for Autonomous Vehicles (MLAV 2017)_, Sydney, Australia. Available at: arXiv:1711.01458 [cs] http://arxiv.org/abs/1711.01458
See https://github.com/SensorsINI/ddd20-itsc20 for the code used for the Hu paper above.
# Installation instructions using conda and Python 2.7
This project currently works with Python 2.7 under linux. Lasted tested working *view.py* November 2024.
*(Don't try to run this code in python 3; it depends on some cryptic multiprocessing code that is not portable to python3! Trust us, we tried to port it.)*
1. First, create an Python 2.7 environment
```bash
conda create -n ddd20 python=2.7
conda activate ddd20
```
2. Install all dependencies:
```bash
pip install future
pip install numpy h5py opencv-python==4.2.0.32 openxc==0.15.0
```
3. There is no step 3, have fun! :tada:
# Usage:
See https://sites.google.com/view/davis-driving-dataset-2020/home for details
## viewing
- Play a file from the beginning
```bash
$ python view.py
```
- Print usage
```bash
$ python view.py --help
usage: view.py [-h] [--start START] [--rotate ROTATE] filename
positional arguments:
filename
optional arguments:
-h, --help show this help message and exit
--start START, -s START
Examples:
-s 50% - play file starting at 50%
-s 66s - play file starting at 66s
--rotate ROTATE, -r ROTATE
Rotate the scene 180 degrees if True, Otherwise False
```
While viewing, hit ? or h for help in console:
```
space pause
b brighter
d darker
s slower
f faster
i toggle/rotate info
r rotate 180 deg
```
## Exporting raw data into standard data types
The DDD20 recordings are recorded using a custom data structure in HDF5.
This design choice made the batch processing restricted without reformatting/exporting.
We prepared a script that can convert the original HDF5 recording into a
nicer data strcture that user can directly work on. __However, this file will not contain the car CAN bus steering/throttle/GPS, etc.__
```bash
$ python export_ddd20_hdf.py [-h] [--rotate ROTATE] filename
```
The newly exported file is an HDF5 file that is called `filename.exported.hdf5`.
This file is saved at the same folder of the `filename`.
This HDF5 file has a very simple structure, it has three datasets:
```
event: (N events x 4) # each row is an event.
frame: (M frames x 260 x 346)
frame_ts: (M frames x 1)
```
Added now is option to turn off the display (thanks youkaichao) so that issue #4 can be resolved by simply adding the option:
```
python export_ddd20_hdf.py filename --display 0
```
## Exporting to frame-based representation
```bash
$ python export.py [-h] [--tstart TSTART] [--tstop TSTOP] [--binsize BINSIZE]
[--update_prog_every UPDATE_PROG_EVERY]
[--export_aps EXPORT_APS] [--export_dvs EXPORT_DVS] [--display 0]
[--out_file OUT_FILE]
filename
```
# License
This software is released under the GNU LESSER GENERAL PUBLIC LICENSE Version 3.