Adds fill paramter to utils.draw_bounding_boxes #3296

oke-aditya · 2021-01-26T09:47:39Z

Closes #3280 .

Currently, I have kept fill to Tuple[int, int, int, int] or str.
Let me know if changes are needed.

Also, let me know how we can test this.

datumbox

Thanks for the PR! I left a couple of comments for your consideration.

Also could you also change the unit-test to support the new param once you finalize the implementation? You will need to produce a new expected file but I think the code already supports this if you delete the old file.

torchvision/utils.py

oke-aditya · 2021-01-27T06:00:34Z

torchvision/utils.py

    txt_font = ImageFont.load_default() if font is None else ImageFont.truetype(font=font, size=font_size)

    for i, bbox in enumerate(img_boxes):
        color = None if colors is None else colors[i]
-        draw.rectangle(bbox, width=width, outline=color)
+        if fill:
+            fill_color = color + [100]


I think we would need to set alpha channel to some hardcoded value. Since that gives the semi transparent look.

I just noticed that this causes issues on the type checks. Also this is not going to work for all values of color. There are various corner cases:

The color could be None

It could be a string

Or a tuple.

It might be worth confirming that all work as expected by adding a test, what you think?

If fill is true

If color is None lets fill with default color.

If color is Tuple[int, int, int] I think we just add alpha of 100 as above and do.

If color is str I think we need to get RGB value of the str add default alpha of 100 and draw ?

I think we need to handle the fill_color = color + [100] especially in each corner case:

>>> None + [100] Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: unsupported operand type(s) for +: 'NoneType' and 'list' >>> (0, 0, 0) + [100] Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: can only concatenate tuple (not "list") to tuple >>> "red" + [100] Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: can only concatenate str (not "list") to str

Sure. Sorry, I'm bit low on bandwidth this week but will complete by next week.
I will add tests too to check these corner cases. It's really simple to handle and I should have thought before 😅

Don't worry mate. :) Take your time! Ping me when ready to make sure I review it again. Thanks!

datumbox

Would you be able to update the following test to use your new feature?

vision/test/test_utils.py

Lines 83 to 97 in a0e5bbe

    
           def test_draw_boxes(self): 
        
               img = torch.full((3, 100, 100), 255, dtype=torch.uint8) 
        
               boxes = torch.tensor([[0, 0, 20, 20], [0, 0, 0, 0], 
        
                                    [10, 15, 30, 35], [23, 35, 93, 95]], dtype=torch.float) 
        
               labels = ["a", "b", "c", "d"] 
        
               colors = ["green", "#FF00FF", (0, 255, 0), "red"] 
        
               result = utils.draw_bounding_boxes(img, boxes, labels=labels, colors=colors) 
        
               path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "assets", "fakedata", "draw_boxes_util.png") 
        
               if not os.path.exists(path): 
        
                   res = Image.fromarray(result.permute(1, 2, 0).contiguous().numpy()) 
        
                   res.save(path) 
        
               expected = torch.as_tensor(np.array(Image.open(path))).permute(2, 0, 1) 
        
               self.assertTrue(torch.equal(result, expected))

Feel free to update the expected file as well.

datumbox

@oke-aditya I added a couple of more comments, please let me know what you think.

datumbox

.

oke-aditya · 2021-01-30T18:01:51Z

Done ! @datumbox It works well now, semi transparent images look good.
I just modified the test and the expected file.
Which should now run with filled option. Since it covers all cases it should be fine.

And sorry for the delay, it was really simple to fix, but I was on vacation 😅

codecov · 2021-01-30T18:33:01Z

Codecov Report

Merging #3296 (6134881) into master (1e3a183) will decrease coverage by 0.06%.
The diff coverage is 43.75%.

@@            Coverage Diff             @@
##           master    #3296      +/-   ##
==========================================
- Coverage   73.96%   73.90%   -0.07%     
==========================================
  Files         104      104              
  Lines        9607     9618      +11     
  Branches     1537     1543       +6     
==========================================
+ Hits         7106     7108       +2     
- Misses       2024     2028       +4     
- Partials      477      482       +5

Impacted Files	Coverage Δ
torchvision/utils.py	`59.57% <43.75%> (-5.49%)`	⬇️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1e3a183...e28d6cc. Read the comment docs.

datumbox

LGTM, thanks a lot for the PR @oke-aditya!

Summary: * adds fill paramter * small doc edit * Change fill to bool * add filled * fix the bugs * Fixes bugs * adds test with fill param * fix tuple bug Reviewed By: datumbox Differential Revision: D26226612 fbshipit-source-id: 485e9aec56c4f92da64c37a080ebcb6a2182a76f Co-authored-by: Vasilis Vryniotis <[email protected]>

adds fill paramter

db45318

facebook-github-bot added the cla signed label Jan 26, 2021

small doc edit

1f1095b

datumbox reviewed Jan 26, 2021

View reviewed changes

torchvision/utils.py Outdated Show resolved Hide resolved

torchvision/utils.py Outdated Show resolved Hide resolved

oke-aditya added 2 commits January 27, 2021 10:03

Change fill to bool

fdf6f76

add filled

978ad4a

oke-aditya commented Jan 27, 2021

View reviewed changes

torchvision/utils.py Outdated Show resolved Hide resolved

fix the bugs

9956bd1

oke-aditya commented Jan 27, 2021

View reviewed changes

datumbox approved these changes Jan 27, 2021

View reviewed changes

datumbox self-requested a review January 27, 2021 14:38

datumbox reviewed Jan 27, 2021

View reviewed changes

datumbox marked this pull request as draft January 27, 2021 14:44

datumbox suggested changes Jan 27, 2021

View reviewed changes

oke-aditya and others added 4 commits January 30, 2021 21:59

Fixes bugs

efe809e

Merge branch 'master' of https://github.com/pytorch/vision into add_fill

430f45c

adds test with fill param

54d44cc

fix tuple bug

6134881

oke-aditya marked this pull request as ready for review January 30, 2021 18:02

oke-aditya requested a review from datumbox January 30, 2021 18:02

datumbox approved these changes Feb 1, 2021

View reviewed changes

Merge branch 'master' into add_fill

e28d6cc

datumbox merged commit b2cf604 into pytorch:master Feb 1, 2021

oke-aditya deleted the add_fill branch February 1, 2021 11:53

oke-aditya mentioned this pull request Mar 22, 2021

Improved utilites, adds examples, tests #3594

Merged

6 tasks

oke-aditya mentioned this pull request Dec 27, 2021

Random colors for drawing boxes #5127

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Adds fill paramter to utils.draw_bounding_boxes #3296

Adds fill paramter to utils.draw_bounding_boxes #3296

Uh oh!

oke-aditya commented Jan 26, 2021

Uh oh!

datumbox left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

oke-aditya Jan 27, 2021

Uh oh!

datumbox Jan 27, 2021

Uh oh!

oke-aditya Jan 27, 2021

Uh oh!

datumbox Jan 27, 2021

Uh oh!

oke-aditya Jan 27, 2021 •

edited

Loading

Uh oh!

datumbox Jan 27, 2021

Uh oh!

datumbox left a comment •

edited

Loading

Uh oh!

datumbox left a comment

Uh oh!

datumbox left a comment

Uh oh!

oke-aditya commented Jan 30, 2021 •

edited

Loading

Uh oh!

codecov bot commented Jan 30, 2021 •

edited

Loading

Uh oh!

datumbox left a comment

Uh oh!

Uh oh!

	def test_draw_boxes(self):
	img = torch.full((3, 100, 100), 255, dtype=torch.uint8)
	boxes = torch.tensor([[0, 0, 20, 20], [0, 0, 0, 0],
	[10, 15, 30, 35], [23, 35, 93, 95]], dtype=torch.float)
	labels = ["a", "b", "c", "d"]
	colors = ["green", "#FF00FF", (0, 255, 0), "red"]
	result = utils.draw_bounding_boxes(img, boxes, labels=labels, colors=colors)

	path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "assets", "fakedata", "draw_boxes_util.png")
	if not os.path.exists(path):
	res = Image.fromarray(result.permute(1, 2, 0).contiguous().numpy())
	res.save(path)

	expected = torch.as_tensor(np.array(Image.open(path))).permute(2, 0, 1)
	self.assertTrue(torch.equal(result, expected))

Adds fill paramter to utils.draw_bounding_boxes #3296

Adds fill paramter to utils.draw_bounding_boxes #3296

Uh oh!

Conversation

oke-aditya commented Jan 26, 2021

Uh oh!

datumbox left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

oke-aditya Jan 27, 2021

Choose a reason for hiding this comment

Uh oh!

datumbox Jan 27, 2021

Choose a reason for hiding this comment

Uh oh!

oke-aditya Jan 27, 2021

Choose a reason for hiding this comment

Uh oh!

datumbox Jan 27, 2021

Choose a reason for hiding this comment

Uh oh!

oke-aditya Jan 27, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

datumbox Jan 27, 2021

Choose a reason for hiding this comment

Uh oh!

datumbox left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

datumbox left a comment

Choose a reason for hiding this comment

Uh oh!

datumbox left a comment

Choose a reason for hiding this comment

Uh oh!

oke-aditya commented Jan 30, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Jan 30, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

datumbox left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

oke-aditya Jan 27, 2021 •

edited

Loading

datumbox left a comment •

edited

Loading

oke-aditya commented Jan 30, 2021 •

edited

Loading

codecov bot commented Jan 30, 2021 •

edited

Loading