python - 如何添加具有所需数量的点的叠加？

Question

我正在尝试训练形状预测器，但遇到了一个问题，即 add_overlay 函数需要 68 个点中的 5 个。那么，如何添加 46 点的叠加层呢？这是代码，它与 example 几乎相同在文档中。

#!/usr/bin/python
import os
import sys
import glob

import dlib
from skimage import io



if len(sys.argv) != 2:
    print(
        "Give the path to the examples/faces directory as the argument to this "
        "program. For example, if you are in the python_examples folder then "
        "execute this program by running:\n"
        "    ./train_shape_predictor.py ../examples/faces")
    exit()
faces_folder = sys.argv[1]

options = dlib.shape_predictor_training_options()

options.oversampling_amount = 500

options.tree_depth = 2
options.be_verbose = True

training_xml_path = os.path.join(faces_folder, "women_test.xml")
dlib.train_shape_predictor(training_xml_path, "predictor.dat", options)

print("\nTraining accuracy: {}".format(
    dlib.test_shape_predictor(training_xml_path, "predictor.dat")))

predictor = dlib.shape_predictor("predictor.dat")
detector = dlib.simple_object_detector("detector.svm")


print("Showing detections and predictions on the images in the objects folder...")
win = dlib.image_window()
for f in glob.glob(os.path.join(faces_folder, "*.jpg")):
    print("Processing file: {}".format(f))
    img = io.imread(f)

    win.clear_overlay()
    win.set_image(img)

    dets = detector(img, 1)
    print("Number of faces detected: {}".format(len(dets)))
    for k, d in enumerate(dets):
        print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
            k, d.left(), d.top(), d.right(), d.bottom()))
        shape = predictor(img, d)
        print("Part 0: {}, Part 1: {} ...".format(shape.part(0),
                                                  shape.part(1)))
        win.add_overlay(shape)

    win.add_overlay(dets)
    dlib.hit_enter_to_continue()

输出日志:

Training with cascade depth: 10
Training with tree depth: 2
Training with 500 trees per cascade level.
Training with nu: 0.05
Training with random seed: 
Training with oversampling amount: 500
Training with feature pool size: 400
Training with feature pool region padding: 0
Training with lambda_param: 0.1
Training with 20 split tests.
Fitting trees...
Training complete                           
Training complete, saved predictor to file predictor.dat

Training accuracy: 0.0
Showing detections and predictions on the images in the faces folder...
Processing file: img/women/women5.jpg
Number of faces detected: 1
Detection 0: Left: 290 Top: 498 Right: 646 Bottom: 676
Part 0: (317, 564), Part 1: (319, 582) ...
Traceback (most recent call last):
  File "train_shape_detector.py", line 131, in <module>
    win.add_overlay(shape)
RuntimeError: 

Error detected at line 25.
Error detected in file /tmp/pip-build-867r6kjx/dlib/dlib/../dlib/image_processing/render_face_detections.h.
Error detected in function std::vector<dlib::image_display::overlay_line> dlib::render_face_detections(const std::vector<dlib::full_object_detection>&, dlib::rgb_pixel).

Failing expression was dets[i].num_parts() == 68 || dets[i].num_parts() == 5.
     std::vector<image_window::overlay_line> render_face_detections()
     You have to give either a 5 point or 68 point face landmarking output to this function. 
     dets[0].num_parts():  46

Answer 1

您正在使用 dlib 窗口，该窗口会检查检测到的点数是 5 还是 68。

就您而言，您有 46 分。您需要在 cv2 窗口上显示图像。

def annotate_landmarks(image, landmarks):
"""
Given image and a set of landmark points, annotates the points for viewing
:param image: Input image
:type image: np.array
:param landmarks: set of facial landmark points
:type landmarks: [(float, float)]
:return: Resulting annotated image
:rtype: np.array
"""
image = image.copy()
for idx, point in enumerate(landmarks):
    pos = (point[0, 0], point[0, 1])
    cv2.putText(image, str(idx), pos,
                fontFace=cv2.FONT_HERSHEY_SCRIPT_SIMPLEX,
                fontScale=0.4,
                color=(0, 0, 255))
    cv2.circle(image, pos, 3, color=(0, 255, 255))
return image

现在使用注释函数来显示结果。

new_img = img
for k, d in enumerate(dets):
    shape = predictor(new_img, d)
    new_img = annotate_landmarks(new_img, shape)

cv2.imshow(new_image)
cv2.waitkey()

注意:此功能现在可能会直接插入您的需求。检查传入 annotate_landmarks 函数的 shape 类型