xiaoxue_admin
/
supervision
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

refactor: 🔄 Rename xyxy_xywh function to xyxy_to_xywh 

docs: 📝 Update docstring for KeyPoints.from_transformers

Signed-off-by: Onuralp SEZER <thunderbirdtr@gmail.com>
This commit is contained in:
Onuralp SEZER 2025-02-18 00:52:27 +03:00
parent fb1dbfd084
commit cbaa605e08
No known key found for this signature in database
GPG Key ID: CF0835DFDF14CA38
4 changed files with 52 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
supervision/__init__.py
View File

@ -78,7 +78,7 @@ from supervision.detection.utils import (
xcycwh_to_xyxy,
xywh_to_xyxy,
xyxy_to_polygons,
xyxy_xywh,
xyxy_to_xywh,
)
from supervision.draw.color import Color, ColorPalette
from supervision.draw.utils import (
@ -227,5 +227,5 @@ __all__ = [
"xcycwh_to_xyxy",
"xywh_to_xyxy",
"xyxy_to_polygons",
"xyxy_xywh",
"xyxy_to_xywh",
]

4
supervision/detection/utils.py
View File

@ -321,7 +321,7 @@ def xywh_to_xyxy(xywh: np.ndarray) -> np.ndarray:
return xyxy
def xyxy_xywh(xyxy: np.ndarray) -> np.ndarray:
def xyxy_to_xywh(xyxy: np.ndarray) -> np.ndarray:
"""
Converts bounding box coordinates from `(x_min, y_min, x_max, y_max)`
format to `(x, y, width, height)` format.
@ -345,7 +345,7 @@ def xyxy_xywh(xyxy: np.ndarray) -> np.ndarray:
[15, 25, 50, 70]
])
sv.xyxy_xywh(xyxy=xyxy)
sv.xyxy_to_xywh(xyxy=xyxy)
# array([
# [10, 20, 30, 40],
# [15, 25, 35, 45]

52
supervision/keypoint/core.py
View File

@ -526,53 +526,49 @@ class KeyPoints:
Example:
```python
import requests
import torch
from PIL import Image
import requests
import supervision as sv
import torch
from transformers import (
AutoProcessor,
RTDetrForObjectDetection,
VitPoseForPoseEstimation,
)
import supervision as sv
device = "cuda" if torch.cuda.is_available() else "cpu"
image = Image.open(<SOURCE_IMAGE_PATH>)
person_image_processor = AutoProcessor.from_pretrained("PekingU/rtdetr_r50vd_coco_o365")
person_model = RTDetrForObjectDetection.from_pretrained("PekingU/rtdetr_r50vd_coco_o365", device_map=device)
DETECTION_MODEL_ID = "PekingU/rtdetr_r50vd_coco_o365"
inputs = person_image_processor(images=image, return_tensors="pt").to(device)
detection_processor = AutoProcessor.from_pretrained(DETECTION_MODEL_ID, use_fast=True)
detection_model = RTDetrForObjectDetection.from_pretrained(DETECTION_MODEL_ID, device_map=DEVICE)
inputs = detection_processor(images=frame, return_tensors="pt").to(DEVICE)
with torch.no_grad():
outputs = person_model(**inputs)
outputs = detection_model(**inputs)
results = person_image_processor.post_process_object_detection(
outputs, target_sizes=torch.tensor([(image.height, image.width)]), threshold=0.3
)
result = results[0] # take first image results
detections = sv.Detections.from_transformers(result)
person_detections_xywh = sv.xyxy_xywh(detections[detections.class_id == 0].xyxy)
target_size = torch.tensor([(frame.height, frame.width)])
results = detection_processor.post_process_object_detection(
outputs, target_sizes=target_size, threshold=0.3)
image_processor = AutoProcessor.from_pretrained("usyd-community/vitpose-base-simple")
model = VitPoseForPoseEstimation.from_pretrained(
"usyd-community/vitpose-base-simple", device_map=device
)
detections = sv.Detections.from_transformers(results[0])
boxes = sv.xyxy_to_xywh(detections[detections.class_id == 0].xyxy)
inputs = image_processor(image, boxes=[person_detections_xywh], return_tensors="pt").to(
device
)
POSE_ESTIMATION_MODEL_ID = "usyd-community/vitpose-base-simple"
pose_estimation_processor = AutoProcessor.from_pretrained(POSE_ESTIMATION_MODEL_ID)
pose_estimation_model = VitPoseForPoseEstimation.from_pretrained(
POSE_ESTIMATION_MODEL_ID, device_map=DEVICE)
inputs = pose_estimation_processor(frame, boxes=[boxes], return_tensors="pt").to(DEVICE)
with torch.no_grad():
outputs = model(**inputs)
pose_results = image_processor.post_process_pose_estimation(
outputs, boxes=[person_detections_xywh]
)[0]
keypoints = sv.KeyPoints.from_transformers(pose_results)
outputs = pose_estimation_model(**inputs)
results = pose_estimation_processor.post_process_pose_estimation(outputs, boxes=[boxes])
key_point = sv.KeyPoints.from_transformers(results[0])
```
""" # noqa: E501 // docs

24
test/detection/test_utils.py
View File

@ -21,6 +21,7 @@ from supervision.detection.utils import (
scale_boxes,
xcycwh_to_xyxy,
xywh_to_xyxy,
xyxy_to_xywh,
)
TEST_MASK = np.zeros((1, 1000, 1000), dtype=bool)
@ -1381,6 +1382,29 @@ def test_xywh_to_xyxy(xywh: np.ndarray, expected_result: np.ndarray) -> None:
np.testing.assert_array_equal(result, expected_result)
@pytest.mark.parametrize(
"xyxy, expected_result",
[
(np.array([[10, 20, 40, 60]]), np.array([[10, 20, 30, 40]])), # standard case
(np.array([[0, 0, 0, 0]]), np.array([[0, 0, 0, 0]])), # zero size bounding box
(
np.array([[50, 50, 150, 150]]),
np.array([[50, 50, 100, 100]]),
), # large bounding box
(
np.array([[-10, -20, 20, 20]]),
np.array([[-10, -20, 30, 40]]),
), # negative coordinates
(np.array([[50, 50, 50, 80]]), np.array([[50, 50, 0, 30]])), # zero width
(np.array([[50, 50, 70, 50]]), np.array([[50, 50, 20, 0]])), # zero height
(np.array([]).reshape(0, 4), np.array([]).reshape(0, 4)), # empty array
],
)
def test_xyxy_to_xywh(xyxy: np.ndarray, expected_result: np.ndarray) -> None:
result = xyxy_to_xywh(xyxy)
np.testing.assert_array_equal(result, expected_result)
@pytest.mark.parametrize(
"xcycwh, expected_result",
[