100~105 HOG

Author: JimmyHHua

Diff for: README.md

@@ -137,4 +137,10 @@ code_095 | [ORB Feature Descriptor Matching](python/code_095)   | ✔️
 code_096 | [Multiple  Descriptor Matching Methods](python/code_096)   | ✏️
 code_097 | [Location of Known Objects Based on Descriptor Matches](python/code_097)   | ✏️
 code_098 | [SIFT Feature Critical Point Detection](python/code_097)   | ✔️
-code_099 | [SIFT Feature Descriptor Matching](python/code_097)   | ✔️
+code_099 | [SIFT Feature Descriptor Matching](python/code_097)   | ✔️
+code_100 | [HOG Pedestrian Detection](python/code_100/opencv_100.py)   | ✔️
+code_101 | [HOG Multiscale Detection](python/code_101/opencv_101.py)   | ✏️
+code_102 | [HOG Extract Descriptor](python/code_102/opencv_102.py)   | ✔️
+code_103 | [HOG Use Descriptors to Generate Sample Data](python/code_103/opencv_103.py)   | ✔️
+code_104 | [(Detection Case)-HOG+SVM Train](python/code_104/opencv_104.py)   | ✔️
+code_105 | [(Detection Case)-HOG+SVM Predict](python/code_105/opencv_105.py)   | ✔️

Diff for: README_CN.md

@@ -136,4 +136,10 @@ code_095 | [ORB 特征描述子匹配](python/code_095)   | ✔️
 code_096 | [多种描述子匹配方法](python/code_096)   | ✏️
 code_097 | [基于描述子匹配的已知对象定位](python/code_097)   | ✏️
 code_098 | [SIFT 特征关键点检测](python/code_097)   | ✔️
-code_099 | [SIFT 特征描述子匹配](python/code_097)   | ✔️
+code_099 | [SIFT 特征描述子匹配](python/code_097)   | ✔️
+code_100 | [HOG 行人检测](python/code_100/opencv_100.py)   | ✔️
+code_101 | [HOG 多尺度检测](python/code_101/opencv_101.py)   | ✏️
+code_102 | [HOG 提取描述子](python/code_102/opencv_102.py)   | ✔️
+code_103 | [HOG 使用描述子生成样本数据](python/code_103/opencv_103.py)   | ✔️
+code_104 | [(检测案例)-HOG+SVM 训练](python/code_104/opencv_104.py)   | ✔️
+code_105 | [(检测案例)-HOG+SVM 预测](python/code_105/opencv_105.py)   | ✔️

Diff for: python/code_100/README.md

@@ -0,0 +1,2 @@
+#### More Detail, please check the blog of zhihu as below
+✈️ ✈️ ✈️ [OpenCV图像处理-HOG特征和应用](https://zhuanlan.zhihu.com/p/75705284)

Diff for: python/code_100/opencv_100.py

@@ -0,0 +1,19 @@
+import cv2 as cv
+
+if __name__ == '__main__':
+    src = cv.imread("pedestrian.png")
+    cv.imshow("input", src)
+    hog = cv.HOGDescriptor()
+    hog.setSVMDetector(cv.HOGDescriptor_getDefaultPeopleDetector())
+    # Detect people in the image
+    (rects, weights) = hog.detectMultiScale(src,
+                                            winStride=(4, 4),
+                                            padding=(8, 8),
+                                            scale=1.25,
+                                            useMeanshiftGrouping=False)
+    for (x, y, w, h) in rects:
+        cv.rectangle(src, (x, y), (x + w, y + h), (0, 255, 0), 2)
+
+    cv.imshow("hog-detector", src)
+    cv.waitKey(0)
+    cv.destroyAllWindows()

Diff for: python/code_100/pedestrian.png

@@ -0,0 +1,2 @@
+#### More Detail, please check the blog of zhihu as below
+✈️ ✈️ ✈️ [OpenCV图像处理-HOG特征和应用](https://zhuanlan.zhihu.com/p/75705284)

Diff for: python/code_101/README.md

@@ -0,0 +1,22 @@
+import cv2 as cv
+
+if __name__ == '__main__':
+    src = cv.imread("people1.jpg")
+    src = cv.resize(src, (0, 0), fx=0.5, fy=0.5)
+    cv.imshow("input", src)
+    cv.imwrite('people_src.jpg', src)
+    hog = cv.HOGDescriptor()
+    hog.setSVMDetector(cv.HOGDescriptor_getDefaultPeopleDetector())
+    # Detect people in the image
+    (rects, weights) = hog.detectMultiScale(src,
+                                            winStride=(4, 4),
+                                            padding=(8, 8),
+                                            scale=1.25,
+                                            useMeanshiftGrouping=False)
+    for (x, y, w, h) in rects:
+        cv.rectangle(src, (x, y), (x + w, y + h), (0, 0, 255), 2)
+
+    cv.imshow("hog-detector", src)
+    cv.imwrite('detector.jpg', src)
+    cv.waitKey(0)
+    cv.destroyAllWindows()

Diff for: python/code_101/detector.jpg

@@ -0,0 +1,2 @@
+#### More Detail, please check the blog of zhihu as below
+✈️ ✈️ ✈️ [OpenCV图像处理-HOG特征和应用](https://zhuanlan.zhihu.com/p/75705284)

Diff for: python/code_101/opencv_101.py

@@ -0,0 +1,15 @@
+import cv2 as cv
+import numpy as np
+
+if __name__ == '__main__':
+    src = cv.imread("gaoyy_min.png")
+    print(src.shape)
+    hog = cv.HOGDescriptor()
+    gray = cv.cvtColor(src, cv.COLOR_BGR2GRAY)
+    fv = hog.compute(gray, winStride=(8, 8), padding=(0, 0))
+    print(len(fv))
+    print(fv[0])
+    cv.namedWindow('hog-descriptor', cv.WINDOW_NORMAL)
+    cv.imshow("hog-descriptor", src)
+    cv.waitKey(0)
+    cv.destroyAllWindows()

Diff for: python/code_101/pedestrian_02.png

@@ -0,0 +1,2 @@
+#### More Detail, please check the blog of zhihu as below
+✈️ ✈️ ✈️ [OpenCV图像处理-HOG特征和应用](https://zhuanlan.zhihu.com/p/75705284)

Diff for: python/code_101/people.jpg

@@ -0,0 +1,44 @@
+import cv2 as cv
+import os
+import numpy as np
+
+
+# 把目标图放在64x128的灰色图片中间，方便计算描述子
+def get_hog_descriptor(image):
+    hog = cv.HOGDescriptor()
+    h, w = image.shape[:2]
+    rate = 64 / w
+    image = cv.resize(image, (64, np.int(rate*h)))
+    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
+    bg = np.zeros((128, 64), dtype=np.uint8)
+    bg[:,:] = 127
+    h, w = gray.shape
+    dy = (128 - h) // 2
+    bg[dy:h+dy,:] = gray
+    fv = hog.compute(bg, winStride=(8, 8), padding=(0, 0))
+    return fv
+
+def get_data(train_data, labels, path, lableType):
+    for file_name in os.listdir(path):
+        img_dir = os.path.join(path, file_name)
+        img = cv.imread(img_dir)
+        hog_desc = get_hog_descriptor(img)
+        one_fv = np.zeros([len(hog_desc)], dtype=np.float32)
+        for i in range(len(hog_desc)):
+            one_fv[i] = hog_desc[i][0]
+        train_data.append(one_fv)
+        labels.append(lableType)
+    return train_data, labels
+
+def get_dataset(pdir, ndir):
+    train_data = []
+    labels = []
+    train_data, labels =  get_data(train_data, labels, pdir, lableType=1)
+    train_data, labels =  get_data(train_data, labels, ndir, lableType=-1)
+
+    return np.array(train_data, dtype=np.float32), np.array(labels, dtype=np.int32)
+
+if __name__ == '__main__':
+    a, b = get_dataset("pdir/", "ndir/")
+    #cv.destroyAllWindows()
+    print(a.shape, b.shape)

Diff for: python/code_101/people1.jpg

@@ -0,0 +1,2 @@
+#### More Detail, please check the blog of zhihu as below
+✈️ ✈️ ✈️ [OpenCV图像处理-HOG特征和应用](https://zhuanlan.zhihu.com/p/75705284)

Diff for: python/code_101/people_src.jpg

@@ -0,0 +1,71 @@
+import cv2 as cv
+import os
+import numpy as np
+
+# 把目标图放在64x128的灰色图片中间，方便计算描述子
+def get_hog_descriptor(image):
+    hog = cv.HOGDescriptor()
+    h, w = image.shape[:2]
+    rate = 64 / w
+    image = cv.resize(image, (64, np.int(rate*h)))
+    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
+    bg = np.zeros((128, 64), dtype=np.uint8)
+    bg[:,:] = 127
+    h, w = gray.shape
+    dy = (128 - h) // 2
+    bg[dy:h+dy,:] = gray
+    fv = hog.compute(bg, winStride=(8, 8), padding=(0, 0))
+    return fv
+
+def get_data(train_data, labels, path, lableType):
+    for file_name in os.listdir(path):
+        img_dir = os.path.join(path, file_name)
+        img = cv.imread(img_dir)
+        hog_desc = get_hog_descriptor(img)
+        one_fv = np.zeros([len(hog_desc)], dtype=np.float32)
+        for i in range(len(hog_desc)):
+            one_fv[i] = hog_desc[i][0]
+        train_data.append(one_fv)
+        labels.append(lableType)
+    return train_data, labels
+
+def get_dataset(pdir, ndir):
+    train_data = []
+    labels = []
+    train_data, labels =  get_data(train_data, labels, pdir, lableType=1)
+    train_data, labels =  get_data(train_data, labels, ndir, lableType=-1)
+
+    return np.array(train_data, dtype=np.float32), np.array(labels, dtype=np.int32)
+
+
+def svm_train(positive_dir, negative_dir):
+    svm = cv.ml.SVM_create()
+    svm.setKernel(cv.ml.SVM_LINEAR)
+    svm.setType(cv.ml.SVM_C_SVC)
+    svm.setC(2.67)
+    svm.setGamma(5.383)
+    trainData, responses = get_dataset(positive_dir, negative_dir)
+    responses = np.reshape(responses, [-1, 1])
+    svm.train(trainData, cv.ml.ROW_SAMPLE, responses)
+    svm.save('svm_data.dat')
+
+
+def elec_detect(image):
+    hog_desc = get_hog_descriptor(test_img)
+    print(len(hog_desc))
+    one_fv = np.zeros([len(hog_desc)], dtype=np.float32)
+    for i in range(len(hog_desc)):
+        one_fv[i] = hog_desc[i][0]
+    one_fv = np.reshape(one_fv, [-1, len(hog_desc)])
+    print(len(one_fv), len(one_fv[0]))
+    svm = cv.ml.SVM_load('svm_data.dat')
+    result = svm.predict(one_fv)[1]
+    print(result)
+
+
+if __name__ == '__main__':
+    #svm_train("D:/vcprojects/dataset/elec_watch/positive/", "D:/vcprojects/dataset/elec_watch/negative/")
+    # cv.waitKey(0)
+    test_img = cv.imread("test.jpg")
+    elec_detect(test_img)
+    #cv.destroyAllWindows()