prev_time = [time.time() for i in range(len(cam_idx))]
# Open3D visualization
vis = o3d.visualization.Visualizer()
vis.create_window(width=640, height=480)
vis.get_render_option().point_size = 5.0
# Load triangulation class
tri = Triangulation(cam_idx=cam_idx, vis=vis,
use_panoptic_dataset=args.use_panoptic_dataset)
# Load mediapipe and display class
if mode=='hand':
for i in range(len(cam_idx)):
pipe[i] = MediaPipeHand(static_image_mode=False, max_num_hands=1)
disp[i] = DisplayHand(draw3d=True, max_num_hands=1, vis=vis)
elif mode=='body':
for i in range(len(cam_idx)):
pipe[i] = MediaPipeBody(static_image_mode=False, model_complexity=1)
disp[i] = DisplayBody(draw3d=True, vis=vis)
elif mode=='holistic':
for i in range(len(cam_idx)):
pipe[i] = MediaPipeHolistic(static_image_mode=False, model_complexity=1)
disp[i] = DisplayHolistic(draw3d=True, vis=vis)
else:
print('Undefined mode only the following modes are available: \n hand / body / holistic')
sys.exit()
while True:
# Loop through video capture
for i, c in enumerate(cap):
import argparse
from utils_display import DisplayHand
from utils_mediapipe import MediaPipeHand
from utils_joint_angle import GestureRecognition
parser = argparse.ArgumentParser()
parser.add_argument('-m', '--mode', default='eval', help='train / eval')
args = parser.parse_args()
mode = args.mode
# Load mediapipe hand class
pipe = MediaPipeHand(static_image_mode=False, max_num_hands=1)
# Load display class
disp = DisplayHand(max_num_hands=1)
# Start video capture
cap = cv2.VideoCapture(0) # By default webcam is index 0
# Load gesture recognition class
gest = GestureRecognition(mode)
counter = 0
while cap.isOpened():
ret, img = cap.read()
if not ret:
break
# Flip image for 3rd person view
img = cv2.flip(img, 1)
'fx': img_width *
0.9, # Approx 0.7w < f < w https://www.learnopencv.com/approximate-focal-length-for-webcams-and-cell-phone-cameras/
'fy': img_width * 0.9,
'cx': img_width * 0.5, # Approx center of image
'cy': img_height * 0.5,
'width': img_width,
'height': img_height,
}
# Load mediapipe and display class
if mode == 'hand':
pipe = MediaPipeHand(static_image_mode=False,
max_num_hands=2,
intrin=intrin)
disp = DisplayHand(draw3d=True,
draw_camera=True,
max_num_hands=2,
intrin=intrin)
elif mode == 'body':
# Note: As of version 0.8.3 3D joint estimation is only available in full body mode
pipe = MediaPipeBody(static_image_mode=False,
model_complexity=1,
intrin=intrin)
disp = DisplayBody(draw3d=True, draw_camera=True, intrin=intrin)
elif mode == 'holistic':
# Note: As of version 0.8.3 3D joint estimation is only available in full body mode
pipe = MediaPipeHolistic(static_image_mode=False,
model_complexity=1,
intrin=intrin)
disp = DisplayHolistic(draw3d=True, draw_camera=True, intrin=intrin)
# log = False
### Input : Live video of 2 hands playing rock paper scissor
### Output: 2D display of hand keypoint
### with gesture classification (rock=fist, paper=five, scissor=three/yeah)
###############################################################################
import cv2
from utils_display import DisplayHand
from utils_mediapipe import MediaPipeHand
from utils_joint_angle import GestureRecognition
# Load mediapipe hand class
pipe = MediaPipeHand(static_image_mode=False, max_num_hands=2)
# Load display class
disp = DisplayHand(max_num_hands=2)
# Start video capture
cap = cv2.VideoCapture(0) # By default webcam is index 0
# Load gesture recognition class
gest = GestureRecognition(mode='eval')
counter = 0
while cap.isOpened():
ret, img = cap.read()
if not ret:
break
# Flip image for 3rd person view
img = cv2.flip(img, 1)
help='Select mode: face_detect / face / hand / body / holistic')
args = parser.parse_args()
mode = args.mode
# Load mediapipe and display class
if mode == 'face_detect':
pipe = MediaPipeFaceDetect(model_selection=0, max_num_faces=5)
disp = DisplayFaceDetect()
elif mode == 'face':
pipe = MediaPipeFace(static_image_mode=False,
max_num_faces=1,
refine_landmarks=True)
disp = DisplayFace(draw3d=True, refine_landmarks=True)
elif mode == 'hand':
pipe = MediaPipeHand(static_image_mode=False, max_num_hands=2)
disp = DisplayHand(draw3d=True, max_num_hands=2)
elif mode == 'body':
pipe = MediaPipeBody(static_image_mode=False, model_complexity=1)
disp = DisplayBody(draw3d=True)
elif mode == 'holistic':
pipe = MediaPipeHolistic(static_image_mode=False,
model_complexity=1,
refine_face_landmarks=True)
disp = DisplayHolistic(draw3d=True, refine_face_landmarks=True)
else:
print(
'Undefined mode only the following modes are available: \nface / hand / body / holistic'
)
sys.exit()
# Start video capture