ADD dnn 131&132

Author: JimmyHHua

README.md

@@ -168,4 +168,6 @@ code_126 | [DNN- Face Detection based on Residual Network](python/code_126)   |
 code_127 | [DNN- Video Face Detection based on Residual Network](python/code_127)   | ✔️
 code_128 | [DNN- Call the Detection Model of Tensorflow](python/code_128)   | ✔️
 code_129 | [DNN- Call the Openpose Implementation Attitude Assessment](python/code_129)   | ✔️
-code_130 | [DNN- Call YOLO Object Detection Network](python/code_130)   | ✔️
+code_130 | [DNN- Call YOLO Object Detection Network](python/code_130)   | ✔️
+code_131 | [DNN- YOLOv3-tiny Real-time Object Detection](python/code_131)   | ✔️
+code_132 | [DNN- Single and Multiple Image Detection](python/code_132)   | ✔️

README_CN.md

@@ -167,4 +167,6 @@ code_126 | [DNN- 基于残差网络的人脸检测](python/code_126)   | ✔️
 code_127 | [DNN- 基于残差网络的视频人脸检测](python/code_127)   | ✔️
 code_128 | [DNN- 调用tensorflow的检测模型](python/code_128)   | ✔️
 code_129 | [DNN- 调用openpose模型实现姿态评估](python/code_129)   | ✔️
-code_130 | [DNN- 调用YOLO对象检测网络](python/code_130)   | ✔️
+code_130 | [DNN- 调用YOLO对象检测网络](python/code_130)   | ✔️
+code_131 | [DNN- YOLOv3-tiny版本实时对象检测](python/code_131)   | ✔️
+code_132 | [DNN- 单张与多张图像的推断](python/code_132)   | ✔️

python/code_131/README.md

@@ -0,0 +1,6 @@
+## OpenCV DNN 支持YOLOv3-tiny版本实时对象检测
+
+
+YOLOv3的模型在CPU上无法做到实时运行，而YOLO作者提供了个YOLOv3版本的精简版对象检测模型，大小只有30MB左右，它可以在CPU上做到实时运行，这个模型就是YOLOv3-tiny模型。
+
+相比YOLOv3，YOLOv3-tiny只有两个输出层，而且权重参数层与参数文件大小都大大的下降，可以在嵌入式设备与前端实时运行。

python/code_131/opencv_131.py

@@ -0,0 +1,82 @@
+import cv2 as cv
+import numpy as np
+
+yolo_tiny_model = "../model/yolov3-tiny-coco/yolov3-tiny.weights";
+yolo_tiny_cfg = "../model/yolov3-tiny-coco/yolov3-tiny.cfg";
+
+# Load names of classes
+classes = None
+with open("../model/yolov3-tiny-coco/object_detection_classes_yolov3.txt", 'rt') as f:
+    classes = f.read().rstrip('\n').split('\n')
+
+# load tensorflow model
+net = cv.dnn.readNetFromDarknet(yolo_tiny_cfg, yolo_tiny_model)
+# set back-end
+net.setPreferableBackend(cv.dnn.DNN_BACKEND_OPENCV)
+net.setPreferableTarget(cv.dnn.DNN_TARGET_CPU)
+
+cap = cv.VideoCapture()
+height = cap.get(cv.CAP_PROP_FRAME_HEIGHT)
+width = cap.get(cv.CAP_PROP_FRAME_WIDTH)
+vw_out = cv.VideoWriter("1.mp4", cv.VideoWriter_fourcc('D', 'I', 'V', 'X'), 10, (np.int(width), np.int(height)), True)
+index = 0
+while True:
+    ret, image = cap.read()
+    if ret is False:
+        break
+    image = cv.flip(image, 1)
+    h, w = image.shape[:2]
+    # 基于多个Region层输出getUnconnectedOutLayersNames
+    blobImage = cv.dnn.blobFromImage(image, 1.0/255.0, (416, 416), None, True, False);
+    outNames = net.getUnconnectedOutLayersNames()
+    net.setInput(blobImage)
+    outs = net.forward(outNames)
+
+    # Put efficiency information.
+    t, _ = net.getPerfProfile()
+    fps = 1000 / (t * 1000.0 / cv.getTickFrequency())
+    label = 'FPS: %.2f' % fps
+    cv.putText(image, label, (0, 15), cv.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0))
+
+    # 绘制检测矩形
+    classIds = []
+    confidences = []
+    boxes = []
+    for out in outs:
+        for detection in out:
+            scores = detection[5:]
+            classId = np.argmax(scores)
+            confidence = scores[classId]
+            # numbers are [center_x, center_y, width, height]
+            if confidence > 0.5:
+                center_x = int(detection[0] * w)
+                center_y = int(detection[1] * h)
+                width = int(detection[2] * w)
+                height = int(detection[3] * h)
+                left = int(center_x - width / 2)
+                top = int(center_y - height / 2)
+                classIds.append(classId)
+                confidences.append(float(confidence))
+                boxes.append([left, top, width, height])
+
+    # 使用非最大抑制
+    indices = cv.dnn.NMSBoxes(boxes, confidences, 0.5, 0.4)
+    for i in indices:
+        i = i[0]
+        box = boxes[i]
+        left = box[0]
+        top = box[1]
+        width = box[2]
+        height = box[3]
+        cv.rectangle(image, (left, top), (left+width, top+height), (0, 0, 255), 2, 8, 0)
+        cv.putText(image, classes[classIds[i]], (left, top),
+                   cv.FONT_HERSHEY_SIMPLEX, 1.0, (255, 255, 0), 2)
+    c = cv.waitKey(1)
+    if c == 27:
+        break
+    index += 1
+    if 100 < index < 200:
+        vw_out.write(image)
+    cv.imshow('YOLOv3-tiny-Detection-Demo', image)
+cv.waitKey(0)
+cv.destroyAllWindows()

python/code_132/0.jpg

@@ -0,0 +1,50 @@
+## OpenCV DNN单张与多张图像的推断
+
+OpenCV DNN中支持单张图像推断，同时还支持分批次方式的图像推断，对应的两个相关API分别为blobFromImage与blobFromImages，它们的返回对象都是一个四维的Mat对象-按照顺序分别为NCHW 其组织方式详解如下：
+
+N --> 表示多张图像
+
+C --> 表示接受输入图像的通道数目
+
+H --> 表示接受输入图像的高度
+
+W --> 表示接受输入图像的宽度
+```python
+cv2.dnn.blobFromImage(
+    image,
+    scalefactor = 1.0,
+    size = Size(),
+    mean = Scalar(),
+    swapRB = false,
+    crop = false,
+    ddepth = CV_32F
+)
+```
+```python
+cv2.dnn.blobFromImages(
+	images,
+	scalefactor = 1.0,
+	size = Size(),
+	mean = Scalar(),
+	swapRB = false,
+	crop = false,
+	ddepth = CV_32F 
+)
+```
+
+参数解释:
+
+- Images表示多张图像,image表示单张图像
+- Scalefactor表示放缩
+- Size表示图像大小
+- Mean表示均值
+- swapRB是否交换通道
+- crop是否剪切
+- ddepth 输出的类型，默认是浮点数格式
+
+
+输出：
+
+<img src='0.jpg'>
+
+<img src='1.jpg'>

python/code_132/1.jpg

@@ -0,0 +1,44 @@
+import cv2 as cv
+import numpy as np
+
+bin_model = "../model/google/bvlc_googlenet.caffemodel"
+protxt = "../model/google/bvlc_googlenet.prototxt"
+
+# Load names of classes
+classes = None
+with open("../model/google/classification_classes_ILSVRC2012.txt", 'rt') as f:
+    classes = f.read().rstrip('\n').split('\n')
+
+# load CNN model
+net = cv.dnn.readNetFromCaffe(protxt, bin_model)
+
+# read input data
+image1 = cv.imread("cats.jpg")
+image2 = cv.imread("car.jpg")
+images = []
+images.append(image1)
+images.append(image2)
+blobs = cv.dnn.blobFromImages(np.asarray(images), 1.0, (224, 224), (104, 117,123), False, crop=False)
+print(blobs.shape)
+
+# Run a model
+net.setInput(blobs)
+out = net.forward()
+# Put efficiency information.
+t, _ = net.getPerfProfile()
+label = 'Inference time: %.2f ms' % (t * 1000.0 / cv.getTickFrequency())
+print(out.shape)
+
+# Get a class with a highest score.
+for i in range(len(out)):
+    classId = np.argmax(out[i])
+    confidence = out[i][classId]
+    cv.putText(images[i], label, (0, 15), cv.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0))
+
+    # Print predicted class.
+    text_label = '%s: %.4f' % (classes[classId] if classes else 'Class #%d' % classId, confidence)
+    cv.putText(images[i], text_label, (50, 50), cv.FONT_HERSHEY_SIMPLEX, 0.75, (0, 0, 255), 2)
+    cv.imshow("googlenet-demo", images[i])
+    cv.imwrite("%d.jpg"%i, images[i])
+    cv.waitKey(0)
+cv.destroyAllWindows()