Add spiral target shape #192
Conversation
github-actions
bot
added
testing
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #192 +/- ##
=======================================
Coverage 98.42% 98.42%
=======================================
Files 43 43
Lines 1775 1781 +6
Branches 358 356 -2
=======================================
+ Hits 1747 1753 +6
Misses 25 25
Partials 3 3
|
There was a problem hiding this comment.
This looks incredible on the soccer dataset – thanks @nickc6424! A few small things to address here.
| Notes | ||
| ----- | ||
| The formula for a spiral can be found here: | ||
| https://en.wikipedia.org/wiki/Archimedean_spiral |
| # Number of rotations | ||
| num_rotations = 3 | ||
|
|
||
| t = np.linspace(0, 1, num=200) |
There was a problem hiding this comment.
Can we add a comment for this one too?
| x = (t * radius) * np.cos(2 * num_rotations * math.pi * t) + cx | ||
| y = (t * radius) * np.sin(2 * num_rotations * math.pi * t) + cy |
There was a problem hiding this comment.
How about we modify this to start at the top when the y range is bigger than the x range? For example, on the dog shape, the outermost point of the spiral would be at the dog's head instead of way off to the right.
| x = (t * radius) * np.cos(2 * num_rotations * math.pi * t) + cx | ||
| y = (t * radius) * np.sin(2 * num_rotations * math.pi * t) + cy |
There was a problem hiding this comment.
| x = (t * radius) * np.cos(2 * num_rotations * math.pi * t) + cx | |
| y = (t * radius) * np.sin(2 * num_rotations * math.pi * t) + cy | |
| x = (t * radius) * np.cos(2 * num_rotations * np.pi * t) + cx | |
| y = (t * radius) * np.sin(2 * num_rotations * np.pi * t) + cy |
| @@ -1,6 +1,7 @@ | |||
| """Shapes that are composed of points.""" | |||
|
|
|||
| import itertools | |||
| import math | |||
There was a problem hiding this comment.
Let's stick with numpy for this.
| import math |
| Class for the spiral shape. | ||
|
|
||
| .. plot:: | ||
| :scale: 100 |
There was a problem hiding this comment.
Let's keep all the plots the same size:
| :scale: 100 | |
| :scale: 75 |
| # Number of rotations | ||
| num_rotations = 3 | ||
|
|
||
| t = np.linspace(0, 1, num=200) |
There was a problem hiding this comment.
Two other thoughts here:
- Can we use fewer points? This would speed up the distance calculations.
- The points are farther apart as the spiral grows. Can we reduce the density at the center and fill in the outer parts better?
Fixes #177
Describe your changes
Added Spiral target shape
Checklist