-
Notifications
You must be signed in to change notification settings - Fork 6.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
draw_keypoints() float support #8276
Changes from 3 commits
7e2a621
d966693
1f2b77c
a6d9f9e
9fa52f0
c9fd6ca
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
Original file line number | Diff line number | Diff line change | ||||
---|---|---|---|---|---|---|
|
@@ -247,6 +247,24 @@ def test_draw_segmentation_masks(colors, alpha, device): | |||||
torch.testing.assert_close(out[:, overlap], interpolated_overlap, rtol=0.0, atol=1.0) | ||||||
|
||||||
|
||||||
def test_draw_keypoints_dtypes(): | ||||||
image_uint8 = torch.full((3, 100, 100), 0, dtype=torch.uint8) | ||||||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. This image should not be just zeros, otherwise it will be easy to miss subtle bugs. This should be the same as for the other test i.e.:
Suggested change
and in fact you'll see that there's a bug because the test will fail |
||||||
image_float = to_dtype(image_uint8, torch.float, scale=True) | ||||||
|
||||||
keypoints_cp = keypoints.clone() | ||||||
|
||||||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. This doesn't seem to be used anywhere
Suggested change
|
||||||
out_uint8 = utils.draw_keypoints(image_uint8, keypoints) | ||||||
out_float = utils.draw_keypoints(image_float, keypoints) | ||||||
|
||||||
assert out_uint8.dtype == torch.uint8 | ||||||
assert out_uint8 is not image_uint8 | ||||||
|
||||||
assert out_float.is_floating_point() | ||||||
assert out_float is not image_float | ||||||
|
||||||
torch.testing.assert_close(out_uint8, to_dtype(out_float, torch.uint8, scale=True), rtol=0, atol=1) | ||||||
|
||||||
|
||||||
def test_draw_segmentation_masks_dtypes(): | ||||||
num_masks, h, w = 2, 100, 100 | ||||||
|
||||||
|
Original file line number | Diff line number | Diff line change |
---|---|---|
|
@@ -336,13 +336,13 @@ def draw_keypoints( | |
|
||
""" | ||
Draws Keypoints on given RGB image. | ||
The values of the input image should be uint8 between 0 and 255. | ||
The image values should be uint8 in [0, 255] or float in [0, 1]. | ||
Keypoints can be drawn for multiple instances at a time. | ||
|
||
This method allows that keypoints and their connectivity are drawn based on the visibility of this keypoint. | ||
|
||
Args: | ||
image (Tensor): Tensor of shape (3, H, W) and dtype uint8. | ||
image (Tensor): Tensor of shape (3, H, W) and dtype uint8 or float. | ||
keypoints (Tensor): Tensor of shape (num_instances, K, 2) the K keypoint locations for each of the N instances, | ||
in the format [x, y]. | ||
connectivity (List[Tuple[int, int]]]): A List of tuple where each tuple contains a pair of keypoints | ||
|
@@ -363,16 +363,16 @@ def draw_keypoints( | |
For more details, see :ref:`draw_keypoints_with_visibility`. | ||
|
||
Returns: | ||
img (Tensor[C, H, W]): Image Tensor of dtype uint8 with keypoints drawn. | ||
img (Tensor[C, H, W]): Image Tensor with keypoints drawn. | ||
""" | ||
|
||
if not torch.jit.is_scripting() and not torch.jit.is_tracing(): | ||
_log_api_usage_once(draw_keypoints) | ||
# validate image | ||
if not isinstance(image, torch.Tensor): | ||
raise TypeError(f"The image must be a tensor, got {type(image)}") | ||
elif image.dtype != torch.uint8: | ||
raise ValueError(f"The image dtype must be uint8, got {image.dtype}") | ||
elif not (image.dtype == torch.uint8 or image.is_floating_point()): | ||
raise ValueError(f"The image dtype must be uint8 or float, got {image.dtype}") | ||
elif image.dim() != 3: | ||
raise ValueError("Pass individual images, not batches") | ||
elif image.size()[0] != 3: | ||
|
@@ -397,6 +397,10 @@ def draw_keypoints( | |
f"Got {visibility.shape = } and {keypoints.shape = }" | ||
) | ||
|
||
original_dtype = image.dtype | ||
if image.is_floating_point(): | ||
image = (image * 255).to(dtype=torch.uint8) | ||
|
||
ndarr = image.permute(1, 2, 0).cpu().numpy() | ||
img_to_draw = Image.fromarray(ndarr) | ||
draw = ImageDraw.Draw(img_to_draw) | ||
|
@@ -428,7 +432,7 @@ def draw_keypoints( | |
width=width, | ||
) | ||
|
||
return torch.from_numpy(np.array(img_to_draw)).permute(2, 0, 1).to(dtype=torch.uint8) | ||
return torch.from_numpy(np.array(img_to_draw)).permute(2, 0, 1).to(dtype=original_dtype) | ||
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Just calling |
||
|
||
|
||
# Flow visualization code adapted from https://github.com/tomrunia/OpticalFlow_Visualization | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's move that test down below, so that it is located next to the rest of the
draw_keypoints
tests. Right now it's in the middle of thedraw_segmentation_mask
tests which is a bit confusing.