增加:MP版本兼容

(被动包含调整复杂系数为1)
This commit is contained in:
leafiber 2022-06-08 15:46:57 +08:00
parent 8feabea649
commit 4a85384e6e
2 changed files with 185 additions and 177 deletions

View File

@ -1,174 +1,177 @@
import cv2 import cv2
import mediapipe as mp import mediapipe as mp
import numpy as np import numpy as np
class HandDetector: class HandDetector:
""" """
使用mediapipe库查找手导出地标像素格式添加了额外的功能 使用mediapipe库查找手导出地标像素格式添加了额外的功能
如查找方式许多手指向上或两个手指之间的距离而且提供找到的手的边界框信息 如查找方式许多手指向上或两个手指之间的距离而且提供找到的手的边界框信息
""" """
def __init__(self, mode=False, max_hands=2, detection_con=0.5, min_track_con=0.5): def __init__(self, mode=False, max_hands=2, detection_con=0.5, min_track_con=0.5):
""" """
:param mode: 在静态模式下对每个图像进行检测 :param mode: 在静态模式下对每个图像进行检测
:param max_hands: 要检测的最大手数 :param max_hands: 要检测的最大手数
:param detection_con: 最小检测置信度 :param detection_con: 最小检测置信度
:param min_track_con: 最小跟踪置信度 :param min_track_con: 最小跟踪置信度
""" """
self.results = None self.results = None
self.mode = mode self.mode = mode
self.max_hands = max_hands self.max_hands = max_hands
self.modelComplex = False self.modelComplex = 1
self.detection_con = detection_con self.detection_con = detection_con
self.min_track_con = min_track_con self.min_track_con = min_track_con
# 初始化手部的识别模型 # 初始化手部的识别模型
self.mpHands = mp.solutions.hands self.mpHands = mp.solutions.hands
self.hands = self.mpHands.Hands(self.mode, self.max_hands, self.detection_con, self.min_track_con) self.hands = self.mpHands.Hands(static_image_mode=self.mode,
self.mpDraw = mp.solutions.drawing_utils # 初始化绘图器 max_num_hands=self.max_hands,
self.tipIds = [4, 8, 12, 16, 20] # 指尖列表 min_detection_confidence=self.detection_con,
self.fingers = [] min_tracking_confidence=self.min_track_con)
self.lmList = [] self.mpDraw = mp.solutions.drawing_utils # 初始化绘图器
self.tipIds = [4, 8, 12, 16, 20] # 指尖列表
def find_hands(self, img, draw=True): self.fingers = []
""" self.lmList = []
从图像(BRG)中找到手部
:param img: 用于查找手的图像 def find_hands(self, img, draw=True):
:param draw: 在图像上绘制输出的标志 """
:return: 带或不带图形的图像 从图像(BRG)中找到手部
""" :param img: 用于查找手的图像
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # 将传入的图像由BGR模式转标准的Opencv模式——RGB模式 :param draw: 在图像上绘制输出的标志
self.results = self.hands.process(img_rgb) :return: 带或不带图形的图像
"""
if self.results.multi_hand_landmarks: img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # 将传入的图像由BGR模式转标准的Opencv模式——RGB模式
for handLms in self.results.multi_hand_landmarks: self.results = self.hands.process(img_rgb)
if draw:
self.mpDraw.draw_landmarks(img, handLms, if self.results.multi_hand_landmarks:
self.mpHands.HAND_CONNECTIONS) for handLms in self.results.multi_hand_landmarks:
return img if draw:
self.mpDraw.draw_landmarks(img, handLms,
def find_position(self, img, hand_no=0, draw=True): self.mpHands.HAND_CONNECTIONS)
""" return img
查找单手的地标并将其放入列表中像素格式还可以返回手部的周围的边界框
:param img: 要查找的主图像 def find_position(self, img, hand_no=0, draw=True):
:param hand_no: 如果检测到多只手则为手部id """
:param draw: 在图像上绘制输出的标志(默认绘制矩形框) 查找单手的地标并将其放入列表中像素格式还可以返回手部的周围的边界框
:return: 像素格式的手部关节位置列表手部边界框 :param img: 要查找的主图像
""" :param hand_no: 如果检测到多只手则为手部id
:param draw: 在图像上绘制输出的标志(默认绘制矩形框)
x_list = [] :return: 像素格式的手部关节位置列表手部边界框
y_list = [] """
onedata = np.zeros([21,3])
zerodata = np.zeros([21,3]) x_list = []
h, w, c = img.shape y_list = []
self.lmList = [] onedata = np.zeros([21,3])
zerodata = np.zeros([21,3])
if self.results.multi_hand_landmarks: h, w, c = img.shape
my_hand = self.results.multi_hand_landmarks[hand_no] self.lmList = []
for i, lm in enumerate(my_hand.landmark):
onedata[i] = np.array([lm.x,lm.y,lm.z]) #将三维坐标添加到单次截屏的数据中 if self.results.multi_hand_landmarks:
my_hand = self.results.multi_hand_landmarks[hand_no]
px, py= int(lm.x * w), int(lm.y * h) for i, lm in enumerate(my_hand.landmark):
x_list.append(px) onedata[i] = np.array([lm.x,lm.y,lm.z]) #将三维坐标添加到单次截屏的数据中
y_list.append(py)
self.lmList.append([px, py]) px, py= int(lm.x * w), int(lm.y * h)
if draw: x_list.append(px)
cv2.circle(img, (px, py), 5, (255, 0, 255), cv2.FILLED) y_list.append(py)
self.lmList.append([px, py])
return onedata, (h, w) if draw:
cv2.circle(img, (px, py), 5, (255, 0, 255), cv2.FILLED)
def fingers_up(self):
""" return onedata, (h, w)
查找列表中打开并返回的手指数会分别考虑左手和右手
:return: 竖起手指的列表 def fingers_up(self):
""" """
fingers = [] 查找列表中打开并返回的手指数会分别考虑左手和右手
if self.results.multi_hand_landmarks: :return: 竖起手指的列表
my_hand_type = self.hand_type() """
# Thumb fingers = []
if my_hand_type == "Right": if self.results.multi_hand_landmarks:
if self.lmList[self.tipIds[0]][0] > self.lmList[self.tipIds[0] - 1][0]: my_hand_type = self.hand_type()
fingers.append(1) # Thumb
else: if my_hand_type == "Right":
fingers.append(0) if self.lmList[self.tipIds[0]][0] > self.lmList[self.tipIds[0] - 1][0]:
else: fingers.append(1)
if self.lmList[self.tipIds[0]][0] < self.lmList[self.tipIds[0] - 1][0]: else:
fingers.append(1) fingers.append(0)
else: else:
fingers.append(0) if self.lmList[self.tipIds[0]][0] < self.lmList[self.tipIds[0] - 1][0]:
# 4 Fingers fingers.append(1)
for i in range(1, 5): else:
if self.lmList[self.tipIds[i]][1] < self.lmList[self.tipIds[i] - 2][1]: fingers.append(0)
fingers.append(1) # 4 Fingers
else: for i in range(1, 5):
fingers.append(0) if self.lmList[self.tipIds[i]][1] < self.lmList[self.tipIds[i] - 2][1]:
return fingers fingers.append(1)
else:
def hand_type(self): fingers.append(0)
""" return fingers
检查传入的手部是左还是右
:return: "Right" "Left" def hand_type(self):
""" """
if self.results.multi_hand_landmarks: 检查传入的手部是左还是右
if self.lmList[17][0] < self.lmList[5][0]: :return: "Right" "Left"
return 1 """
else: if self.results.multi_hand_landmarks:
return 0 if self.lmList[17][0] < self.lmList[5][0]:
return 1
else:
class Main: return 0
def __init__(self, label, N = 100):
self.detector = None
self.camera = cv2.VideoCapture(0, cv2.CAP_DSHOW) class Main:
self.camera.set(3, 1280) def __init__(self, label, N = 100):
self.camera.set(4, 720) self.detector = None
self.N = N self.camera = cv2.VideoCapture(0, cv2.CAP_DSHOW)
#初始化数据包 self.camera.set(3, 1280)
self.label = label self.camera.set(4, 720)
self.data = np.zeros([N,21,3]) self.N = N
self.shape = np.zeros([N,2], dtype = np.int16) #初始化数据包
self.handtype = np.zeros(N, dtype = np.int8) self.label = label
self.data = np.zeros([N,21,3])
def gesture_recognition(self): self.shape = np.zeros([N,2], dtype = np.int16)
self.detector = HandDetector() self.handtype = np.zeros(N, dtype = np.int8)
#初始化数据
def gesture_recognition(self):
zerodata = np.zeros([21,3]) self.detector = HandDetector()
rezult = np.zeros([21,3]) #初始化数据
count = 0
zerodata = np.zeros([21,3])
while True: rezult = np.zeros([21,3])
frame, img = self.camera.read() count = 0
img = self.detector.find_hands(img)
while True:
rezult,shape = self.detector.find_position(img) frame, img = self.camera.read()
if rezult.all() != zerodata.all(): #假设矩阵不为0即捕捉到手部时 img = self.detector.find_hands(img)
self.data[count] = rezult
self.handtype[count] = self.detector.hand_type() rezult,shape = self.detector.find_position(img)
self.shape[count] = np.array(shape) if rezult.all() != zerodata.all(): #假设矩阵不为0即捕捉到手部时
count += 1 self.data[count] = rezult
self.handtype[count] = self.detector.hand_type()
cv2.imshow("camera", img) self.shape[count] = np.array(shape)
key = cv2.waitKey(1) count += 1
if cv2.getWindowProperty('camera', cv2.WND_PROP_VISIBLE) < 1:
break cv2.imshow("camera", img)
elif key == 27: key = cv2.waitKey(1)
break if cv2.getWindowProperty('camera', cv2.WND_PROP_VISIBLE) < 1:
elif count == self.N - 1: break
break elif key == 27:
break
np.savez('firstdata', label = self.label, data = self.data, elif count == self.N - 1:
handtype = self.handtype, shape = self.shape) break
np.savez('firstdata', label = self.label, data = self.data,
if __name__ == '__main__': handtype = self.handtype, shape = self.shape)
Solution = Main(label = "five")
Solution.gesture_recognition()
npzfile = np.load('firstdata.npz') if __name__ == '__main__':
Solution = Main(label = "five")
#print(npzfile['data'][0]) Solution.gesture_recognition()
#print(" ") npzfile = np.load('firstdata.npz')
#print(npzfile['handtype'])
#print(npzfile['label']) #print(npzfile['data'][0])
#print(npzfile['shape']) #print(" ")
#print(npzfile['handtype'])
#print(npzfile['label'])
#print(npzfile['shape'])

11
demo.py
View File

@ -27,14 +27,16 @@ class HandDetector:
self.results = None self.results = None
self.mode = mode self.mode = mode
self.max_hands = max_hands self.max_hands = max_hands
self.modelComplex = False self.modelComplex = 1
self.detection_con = detection_con self.detection_con = detection_con
self.min_track_con = min_track_con self.min_track_con = min_track_con
# 初始化手部的识别模型 # 初始化手部的识别模型
self.mpHands = mp.solutions.hands self.mpHands = mp.solutions.hands
self.hands = self.mpHands.Hands(self.mode, self.max_hands, self.modelComplex, self.hands = self.mpHands.Hands(static_image_mode=self.mode,
self.detection_con, self.min_track_con) max_num_hands=self.max_hands,
min_detection_confidence=self.detection_con,
min_tracking_confidence=self.min_track_con)
self.mpDraw = mp.solutions.drawing_utils # 初始化绘图器 self.mpDraw = mp.solutions.drawing_utils # 初始化绘图器
self.tipIds = [4, 8, 12, 16, 20] # 指尖列表 self.tipIds = [4, 8, 12, 16, 20] # 指尖列表
self.fingers = [] self.fingers = []
@ -154,6 +156,9 @@ class Main:
if (x2 == 1 and x3 == 1) and (x4 == 0 and x5 == 0 and x1 == 0): if (x2 == 1 and x3 == 1) and (x4 == 0 and x5 == 0 and x1 == 0):
cv2.putText(img, "2_TWO", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3, cv2.putText(img, "2_TWO", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
(0, 0, 255), 3) (0, 0, 255), 3)
elif x3 and x1 == 0 and x2 == 0 and (x4 == 0, x5 == 0):
cv2.putText(img, "FUCK YOU!!", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
(0, 0, 255), 3)
elif (x2 == 1 and x3 == 1 and x4 == 1) and (x1 == 0 and x5 == 0): elif (x2 == 1 and x3 == 1 and x4 == 1) and (x1 == 0 and x5 == 0):
cv2.putText(img, "3_THREE", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3, cv2.putText(img, "3_THREE", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
(0, 0, 255), 3) (0, 0, 255), 3)