介绍
YOLO是基于深度学习端到端的实时目标检测系统，YOLO将目标区域预测和目标类别预测整合于单个神经网络模型中，实现在准确率较高的情况下快速目标检测与识别，更加适合现场应用环境。本案例，我们快速实现一个视频目标检测功能，实现的具体原理我们将在单独的文章中详细介绍。
下载编译
我们首先下载Darknet开发框架，Darknet开发框架是YOLO大神级作者自己用C语言编写的开发框架，支持GPU加速，有两种下载方式：
git clone https://github.com/pjreddie/darknet
下载后，完整的文件内容，如下图所示：
编译：
cd darknet
# 编译
make
编译后的文件内容，如下图所示：
下载权重文件
我们这里下载的是“yolov3”版本，大小是200多M，“yolov3-tiny”比较小，30多M。
wget https://pjreddie.com/media/files/yolov3.weights
下载权重文件后，文件内容如下图所示：
上图中的“yolov3-tiny.weights”,"yolov2-tiny.weights"是我单独另下载的。
C语言预测
./darknet detect cfg/yolov3.cfg yolov3.weights data/dog.jpg
如图所示，我们已经预测出三种类别以及对应的概率值。模型输出的照片位于darknet根目录，名字是“predictions.jpg”，如下图所示：
让我们打开模型输出照片看下：
Python语言预测
我们首先需要将“darknet”文件夹内的“libdarknet.so”文件移动到“darknet/python”内，完成后如下图所示：
我们将使用Darknet内置的“darknet.py”,进行预测。预测之前，我们需要对文件进行修改：
默认py文件基于python2.0，所以对于python3.0及以上需要修改print
由于涉及到python和C之间的传值，所以字符串内容需要转码
使用绝对路径
修改完成后，如下图所示：
打开“darknet/cfg/coco.data”文件，将“names”也改为绝对路径(截图内没有修改，读者根据自己的实际路径修改)：
我们可以开始预测了，首先进入“darknet/python”然后执行“darknet.py”文件即可：
结果如下图所示：
对模型输出的结果做个简单的说明，如：
# 分别是：类别，识别概率，识别物体的X坐标，识别物体的Y坐标，识别物体的长度，识别物体的高度
(b'dog', 0.999338686466217, (224.18377685546875, 378.4237060546875, 178.60214233398438, 328.1665954589844)
视频检测
from ctypes import *
import random
import cv2
import numpy as np
def sample(probs):
s = sum(probs)
probs = [a/s for a in probs]
r = random.uniform(0, 1)
for i in range(len(probs)):
r = r - probs[i]
if r <= 0:
return i
return len(probs)-1
def c_array(ctype, values):
arr = (ctype*len(values))()
arr[:] = values
return arr
class BOX(Structure):
_fields_ = [("x", c_float),
("y", c_float),
("w", c_float),
("h", c_float)]
class DETECTION(Structure):
_fields_ = [("bbox", BOX),
("classes", c_int),
("prob", POINTER(c_float)),
("mask", POINTER(c_float)),
("objectness", c_float),
("sort_class", c_int)]
class IMAGE(Structure):
_fields_ = [("w", c_int),
("h", c_int),
("c", c_int),
("data", POINTER(c_float))]
class METADATA(Structure):
_fields_ = [("classes", c_int),
("names", POINTER(c_char_p))]
lib = CDLL("../python/libdarknet.so", RTLD_GLOBAL)
lib.network_width.argtypes = [c_void_p]
lib.network_width.restype = c_int
lib.network_height.argtypes = [c_void_p]
lib.network_height.restype = c_int
predict = lib.network_predict
predict.argtypes = [c_void_p, POINTER(c_float)]
predict.restype = POINTER(c_float)
set_gpu = lib.cuda_set_device
set_gpu.argtypes = [c_int]
make_image = lib.make_image
make_image.argtypes = [c_int, c_int, c_int]
make_image.restype = IMAGE
get_network_boxes = lib.get_network_boxes
get_network_boxes.argtypes = [c_void_p, c_int, c_int, c_float, c_float, POINTER(c_int), c_int, POINTER(c_int)]
get_network_boxes.restype = POINTER(DETECTION)
make_network_boxes = lib.make_network_boxes
make_network_boxes.argtypes = [c_void_p]
make_network_boxes.restype = POINTER(DETECTION)
free_detections = lib.free_detections
free_detections.argtypes = [POINTER(DETECTION), c_int]
free_ptrs = lib.free_ptrs
free_ptrs.argtypes = [POINTER(c_void_p), c_int]
network_predict = lib.network_predict
network_predict.argtypes = [c_void_p, POINTER(c_float)]
reset_rnn = lib.reset_rnn
reset_rnn.argtypes = [c_void_p]
load_net = lib.load_network
load_net.argtypes = [c_char_p, c_char_p, c_int]
load_net.restype = c_void_p
do_nms_obj = lib.do_nms_obj
do_nms_obj.argtypes = [POINTER(DETECTION), c_int, c_int, c_float]
do_nms_sort = lib.do_nms_sort
do_nms_sort.argtypes = [POINTER(DETECTION), c_int, c_int, c_float]
free_image = lib.free_image
free_image.argtypes = [IMAGE]
letterbox_image = lib.letterbox_image
letterbox_image.argtypes = [IMAGE, c_int, c_int]
letterbox_image.restype = IMAGE
load_meta = lib.get_metadata
lib.get_metadata.argtypes = [c_char_p]
lib.get_metadata.restype = METADATA
load_image = lib.load_image_color
load_image.argtypes = [c_char_p, c_int, c_int]
load_image.restype = IMAGE
rgbgr_image = lib.rgbgr_image
rgbgr_image.argtypes = [IMAGE]
predict_image = lib.network_predict_image
predict_image.argtypes = [c_void_p, IMAGE]
predict_image.restype = POINTER(c_float)
def convertBack(x, y, w, h):
xmin = int(round(x - (w / 2)))
xmax = int(round(x + (w / 2)))
ymin = int(round(y - (h / 2)))
ymax = int(round(y + (h / 2)))
return xmin, ymin, xmax, ymax
def array_to_image(arr):
# need to return old values to avoid python freeing memory
arr = arr.transpose(2,0,1)
c, h, w = arr.shape[0:3]
arr = np.ascontiguousarray(arr.flat, dtype=np.float32) / 255.0
data = arr.ctypes.data_as(POINTER(c_float))
im = IMAGE(w,h,c,data)
return im, arr
def detect(net, meta, image, thresh=.5, hier_thresh=.5, nms=.45):
im, image = array_to_image(image)
rgbgr_image(im)
num = c_int(0)
pnum = pointer(num)
predict_image(net, im)
dets = get_network_boxes(net, im.w, im.h, thresh,
hier_thresh, None, 0, pnum)
num = pnum[0]
if nms: do_nms_obj(dets, num, meta.classes, nms)
res = []
for j in range(num):
a = dets[j].prob[0:meta.classes]
if any(a):
ai = np.array(a).nonzero()[0]
for i in ai:
b = dets[j].bbox
res.append((meta.names[i], dets[j].prob[i],
(b.x, b.y, b.w, b.h)))
res = sorted(res, key=lambda x: -x[1])
if isinstance(image, bytes): free_image(im)
free_detections(dets, num)
return res
if __name__ == "__main__":
cap = cv2.VideoCapture(0)
ret, img = cap.read()
fps = cap.get(cv2.CAP_PROP_FPS)
net = load_net(b"/Users/xiaomingtai/darknet/cfg/yolov2-tiny.cfg", b"/Users/xiaomingtai/darknet/yolov2-tiny.weights", 0)
meta = load_meta(b"/Users/xiaomingtai/darknet/cfg/coco.data")
cv2.namedWindow("img", cv2.WINDOW_NORMAL)
while(True):
ret, img = cap.read()
if ret:
r = detect(net, meta, img)
for i in r:
x, y, w, h = i[2][0], i[2][17], i[2][18], i[2][19]
xmin, ymin, xmax, ymax = convertBack(float(x), float(y), float(w), float(h))
pt1 = (xmin, ymin)
pt2 = (xmax, ymax)
cv2.rectangle(img, pt1, pt2, (0, 255, 0), 2)
cv2.putText(img, i[0].decode() + " [" + str(round(i[1] * 100, 2)) + "]", (pt1[0], pt1[1] + 20), cv2.FONT_HERSHEY_SIMPLEX, 1, [0, 255, 0], 4)
cv2.imshow("img", img)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
模型输出结果：
模型视频检测结果：
没有GPU的条件下还是不要选择yolov3了，很慢。
总结
本篇文章主要是YOLO快速上手，我们通过很少的代码就能实现不错的目标检测。当然，想熟练掌握YOLO，理解背后的原理是十分必要的，下篇文章将会重点介绍YOLO原理。