senas/classify.py at master · cazdlt/senas · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
import numpy as np
from matplotlib import pyplot as plt
import argparse
import sys

parser = argparse.ArgumentParser()
parser.add_argument("seña", help="número de la seña en ./señas",
                    type=int)
parser.add_argument("--ignore","-i", action='store_true',help="ignora los videos en los que no el programa funciona correctamente")
args = parser.parse_args()

#número de seña a analizar
fileClassify=int(args.seña)

#Ignora seña si está en la lista de los que no funcionan
fileIgnore=(4,8,9,10,11,14,16,20,21,22,23,24,25,26,30,31,32,33,34,35,36,37,39,42,44,45,47)
if args.ignore and (fileNumber in fileIgnore):
    sys.exit("File ignored")

#seña y (altura, centro, distancia mínima, promedio, distancia recorrida)

def FindDistance(A,B):
 return np.sqrt(np.power((A[0]-B[0]),2) + np.power((A[1]-B[1]),2))

def getInterestFromFile(fileNumber):
    with open('./out2/class%d.txt'%fileNumber) as file:
        name=file.readline()
        line=file.readline()
        hLeft=np.array([float(n) for n in ((line.split('[')[1].split(']')[0]).split())])
        hRight=np.array([float(n) for n in ((line.split('[')[2].split(']')[0]).split())])
        line=file.readline()
        centerLeft=np.array([float(n) for n in ((line.split('[')[1].split(']')[0]).split())])
        centerRight=np.array([float(n) for n in ((line.split('[')[2].split(']')[0]).split())])
        line=file.readline()
        dMinLeft=np.array([float(n) for n in ((line.split('[')[1].split(']')[0]).split())])
        dMinRight=np.array([float(n) for n in ((line.split('[')[2].split(']')[0]).split())])
        line=file.readline()
        avLeft=np.array([float(n) for n in ((line.split('[')[1].split(']')[0]).split())])
        avRight=np.array([float(n) for n in ((line.split('[')[2].split(']')[0]).split())])
        line=file.readline()
        distLeft=np.array([float(n) for n in ((line.split('[')[1].split(']')[0]).split())])
        distRight=np.array([float(n) for n in ((line.split('[')[2].split(']')[0]).split())])

    return name,hLeft,hRight,centerLeft,centerRight,dMinLeft,dMinRight,avLeft,avRight,distLeft,distRight

#distancia de un punto a la cara
def distFace(A):
 return np.sqrt(np.power(A[0],2) + np.power(A[1],2))

#Extrae las trayectorias del archivo
def getTrajectory(fileNumber):
    with open('./out2/traj%d.txt'%fileNumber) as file:
        line=file.readline()
        face=tuple([int(n) for n in (line.split()[0][1:-1],line.split()[1][0:-1])])
        line=file.readline()
        trajLeft=np.array([[int(n) for n in ((line.split('[')[1].split(']')[0]).split())]])
        trajRight=np.array([[int(n) for n in ((line.split('[')[2].split(']')[0]).split())]])
        for line in file:
            trajLeft = np.append(trajLeft, [[int(n) for n in ((line.split('[')[1].split(']')[0]).split())]], 0)
            trajRight = np.append(trajRight, [[int(n) for n in ((line.split('[')[2].split(']')[0]).split())]], 0)

    #cambia referencia de coordenadas en video a  posicion de la cara
    trajLeft[:,0]=trajLeft[:,0]-face[0]
    trajLeft[:,1]=-(trajLeft[:,1]-face[1])
    trajRight[:,0]=trajRight[:,0]-face[0]
    trajRight[:,1]=-(trajRight[:,1]-face[1])
    return trajLeft,trajRight

#extrae puntos de interés a partir de trayectorias de ambas manos
def getInterest(fileNumber):
    trajLeft,trajRight=getTrajectory(fileNumber)
    #Centros de las trayectorias
    xCenterLeft=(max(trajLeft[:,0])-min(trajLeft[:,0]))/2+min(trajLeft[:,0])
    yCenterLeft=(max(trajLeft[:,1])-min(trajLeft[:,1]))/2+min(trajLeft[:,1])
    centerLeft=np.array([xCenterLeft,yCenterLeft])
    xCenterRight=(max(trajRight[:,0])+min(trajRight[:,0]))/2
    yCenterRight=(max(trajRight[:,1])+min(trajRight[:,1]))/2
    centerRight=np.array([xCenterRight,yCenterRight])

    #máxima altura de la trayectoria
    hLeft=trajLeft[np.where(trajLeft[:,0] == min(trajLeft[:,0]))]
    hRight=trajRight[np.where(trajRight[:,0] == min(trajRight[:,0]))]
    hLeft=np.array(hLeft[0])
    hRight=np.array(hRight[0])

    #distancia mínima a la cara
    minLeft=0
    dmin=100000
    for p in trajLeft:
        if abs(distFace(p))<dmin:
            minLeft=p
            dmin=abs(distFace(p))
    minRight=0
    dmin=100000
    for p in trajRight:
        if abs(distFace(p))<dmin:
            minRight=p
            dmin=abs(distFace(p))


    #promedio de las trayectorias
    avxLeft=sum(trajLeft[:,0])/len(trajLeft[:,0])
    avyLeft=sum(trajLeft[:,1])/len(trajLeft[:,1])
    avxRight=sum(trajRight[:,0])/len(trajRight[:,0])
    avyRight=sum(trajRight[:,1])/len(trajRight[:,1])
    avLeft=np.array([avxLeft,avyRight])
    avRight=np.array([avxRight,avyRight])

    #velocidad de las manos
    spxLeft=[(a-b) for a, b in zip(trajLeft[::1,0], trajLeft[1::1,0])]
    spyLeft=[(a-b) for a, b in zip(trajLeft[::1,1], trajLeft[1::1,1])]
    spxRight=[(a-b) for a, b in zip(trajRight[::1,0], trajRight[1::1,0])]
    spyRight=[(a-b) for a, b in zip(trajRight[::1,1], trajRight[1::1,1])]

    #distancia recorrida por las manos
    distLeft=np.array([sum(abs(x) for x in spxLeft),sum(abs(x) for x in spyLeft)])
    distRight=np.array([sum(abs(x) for x in spxRight),sum(abs(x) for x in spyRight)])

    return hLeft,hRight,centerLeft,centerRight,minLeft,minRight,avLeft,avRight,distLeft,distRight

#params=hL,hR,centerL,centerR,dMinL,dMinR,avL,avR,distL,distR
#llena arreglo con parámetros a partir de archivos de entrenamiento
params=np.array([getInterestFromFile(1)[1:]])
names=[getInterestFromFile(1)[0]]
for i,a in enumerate([15,7,18,38,40,3,6,13,43,49,51]):
    params=np.append(params,[getInterestFromFile(a)[1:]],0)
    names.append(getInterestFromFile(a)[0])
seña=np.array(getInterest(fileClassify))

compAll=np.zeros(12)
comp=np.zeros((10, 1))
for i in range(len(params)):
    for j in range(10):
        comp[j]=FindDistance(params[i][j],seña[j])
    compAll[i]=sum(comp)
#print(compAll)
#print(names)
index=np.where(compAll == min(compAll))
guess=names[int(index[0])]
sys.exit(guess)