VSim is a differential motion simulator designed to create simple toy videos to test with potential solutions for the ego-motion problem. Using VSim you can create videos of a square moving in a custom path created by your mouse on a constantly moving background.
input.mp4
input_REF.mp4
Clone this repository. Then run pip install -r requirements.txt
to install.
If you want to convert these videos to DVS, install v2e sim by cloning the
v2e repository as well
Important
VSim requires ffmpeg to be installed on your system
Parameter | Description | Implemented |
---|---|---|
save_dir |
Path to where you want to save the test data | Yes |
ref_save_dir |
Path to where you want to save the reference test data | Yes |
height |
Height of the test data (in pixels) | Yes |
width |
Width of the test data (in pixels) | Yes |
frames |
Number of frames | Yes |
fps |
Frames per second of the test data | Yes |
col_width |
Width of the columns in the moving background (in pixels) | Yes |
object_radius |
Radius of the moving object (in pixels) | Yes |
- Add the parameters to your
params.yaml
filesave_dir
is the path to where you want to save the test data USE .mp4 or it will not workref_save_dir
is the path to where you want to save the reference test data (for use as "ground truth" in the future) USE .mp4 or it will not workheight
is the height of the test data (in pixels)width
is the width of the test data (in pixels)fps
is the frames per second of the test datacol_width
is the width of the columns in the moving background (in pixels)object_radius
is the radius of the moving object (in pixels)fill
is the amount of fill (-1 is filled, 0 is border) > 0 is thickness
- call
create_data.create_data("path_to_params.yaml")
to create the test data (This is in manual path mode, so you will have to manually create the path with your mouse)- You can enter a discrete path instead of the create-your-own by assembling your own function and following the steps in
test.py
- You can enter a discrete path instead of the create-your-own by assembling your own function and following the steps in
Written by Arya Lohia for the IRIS project at Parsa Research Lab at GMU
MIT License