python-eyetracking/tutorial/run_example.py at main · juno-hwang/python-eyetracking · GitHub

History

104 lines (88 loc) · 3.91 KB

Raw

100

101

102

103

104

import time, pickle

import cv2

import mediapipe as mp

import numpy as np

import pandas as pd

from pynput.mouse import Listener

import matplotlib.pyplot as plt

import seaborn as sns

from sklearn.linear_model import LinearRegression

from sklearn.model_selection import train_test_split

from sklearn.gaussian_process import GaussianProcessRegressor

from sklearn.gaussian_process.kernels import RationalQuadratic, RBF, WhiteKernel, DotProduct, ConstantKernel

records = pickle.load(open('records_sample.pkl', 'rb'))

X = [ r['landmark'] for r in records ]

X = np.array(X).reshape(-1, 478*3)

y = [ r['coord'] for r in records ]

y = np.array(y)

print(X.shape, y.shape)

def predict_landmark(landmark):

X = np.array([ [d.x, d.y, d.z] for d in landmark ])

X = X.reshape(1, 478*3)

return model.predict(X)[0]

def predict_landmark_smooth(landmark):

if len(coords_traj) > 0:

prev = coords_traj[-1]

curr = predict_landmark(landmark)

curr = prev*0.8 + curr*0.2

else:

curr = predict_landmark(landmark)

return curr

mp_drawing = mp.solutions.drawing_utils

mp_drawing_styles = mp.solutions.drawing_styles

mp_face_mesh = mp.solutions.face_mesh

drawing_spec = mp_drawing.DrawingSpec(thickness=1, circle_radius=1)

cap = cv2.VideoCapture(0, cv2.CAP_DSHOW)

model = GaussianProcessRegressor(kernel=RationalQuadratic())

model.fit(X, y)

# 웹캠 화면이 최상단에 뜨게 하기 위한 설정

# 최상단을 유지하려는 경우에만 주석을 해제해주세요

# cv2.namedWindow('window', cv2.WINDOW_AUTOSIZE)

# cv2.setWindowProperty('window', cv2.WND_PROP_TOPMOST, 1)

coords_traj = []

with mp_face_mesh.FaceMesh(

max_num_faces=1,

refine_landmarks=True,

min_detection_confidence=0.5,

min_tracking_confidence=0.5) as face_mesh:

while cap.isOpened():

success, image = cap.read()

if not success:

print("웹캠을 찾을 수 없습니다.")

break

image.flags.writeable = False # 성능 향상을 위해 이미지를 읽기 전용으로 만듭니다.

image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

results = face_mesh.process(image)

image.flags.writeable = True

image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)

if results.multi_face_landmarks:

for face_landmarks in results.multi_face_landmarks:

mp_drawing.draw_landmarks(

image=image,

landmark_list=face_landmarks,

connections=mp_face_mesh.FACEMESH_TESSELATION,

landmark_drawing_spec=None,

connection_drawing_spec=mp_drawing_styles

.get_default_face_mesh_tesselation_style())

mp_drawing.draw_landmarks(

image=image,

landmark_list=face_landmarks,

connections=mp_face_mesh.FACEMESH_CONTOURS,

landmark_drawing_spec=None,

connection_drawing_spec=mp_drawing_styles

.get_default_face_mesh_contours_style())

mp_drawing.draw_landmarks(

image=image,

landmark_list=face_landmarks,

connections=mp_face_mesh.FACEMESH_IRISES,

landmark_drawing_spec=None,

connection_drawing_spec=mp_drawing_styles

.get_default_face_mesh_iris_connections_style())

image = cv2.flip(image, 1)

coord = predict_landmark_smooth(face_landmarks.landmark)

coords_traj.append(coord)

cv2.putText(image, f'x: {coord[0]:.2f}, y: {coord[1]:.2f}', (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)

cv2.imshow('window', image)

# 이미지 재생을 위한 대기시간

cv2.waitKey(5)

cap.release()

python-eyetracking/tutorial/run_example.py at main · juno-hwang/python-eyetracking · GitHub

Expand file tree

Breadcrumbs

run_example.py

Latest commit

History

Breadcrumbs

run_example.py

File metadata and controls