def get_landmarks_from_image(self, image_file):
frame = cv2.imread(image_file)
img1, img2, scale, pad = resize_pad(frame)
normalized_palm_detections = self.palm_detector.predict_on_image(img1)
palm_detections = denormalize_detections(normalized_palm_detections,
scale, pad)
xc, yc, scale, theta = self.palm_detector.detection2roi(
palm_detections)
img, affine2, box2 = self.hand_regressor.extract_roi(
frame, xc, yc, theta, scale)
flags2, handed2, normalized_landmarks2 = self.hand_regressor(img)
landmarks2 = self.hand_regressor.denormalize_landmarks(
normalized_landmarks2, affine2)
for i in range(len(flags2)):
landmark, flag = landmarks2[i], flags2[i]
print(flag)
if flag > 0.5:
draw_landmarks(frame,
landmark[:, :2],
HAND_CONNECTIONS,
size=2)
return flags2, handed2, normalized_landmarks2
def test_my_map(map_file_name):
world, path = execute.load_map_from_json(map_file_name)
if not (type(path) == type(list()) and len(path) > 0):
raise ValueError("Path must be at least length 1")
fig, ax = plt.subplots()
ax.set_xlim([0, world.width])
ax.set_ylim([0, world.height])
visualization.draw_landmarks(ax, world.landmarks)
visualization.draw_path(path)
Q_robot = visualization.draw_robot(world.robot_pose)
plt.show()
green_test_message()
face_detections = denormalize_detections(normalized_face_detections, scale,
pad)
xc, yc, scale, theta = face_detector.detection2roi(face_detections.cpu())
img, affine, box = face_regressor.extract_roi(frame, xc, yc, theta, scale)
flags, normalized_landmarks = face_regressor(img.to(gpu))
landmarks = face_regressor.denormalize_landmarks(
normalized_landmarks.cpu(), affine)
#frame = cv2.rectangle(frame, (0,0), (400,400), (120, 120, 120), 400)
for i in range(len(flags)):
landmark, flag = landmarks[i], flags[i]
if flag > .5:
draw_landmarks(frame, landmark[:, :2], FACE_CONNECTIONS, size=1)
draw_roi(frame, box)
draw_detections(frame, face_detections)
cv2.imshow(WINDOW, frame[:, :, :])
# cv2.imwrite('sample/%04d.jpg'%frame_ct, frame[:,:,::-1])
hasFrame, frame = capture.read()
key = cv2.waitKey(1)
if key == 27:
break
capture.release()
cv2.destroyAllWindows()
img1, img2, scale, pad = resize_pad(frame)
normalized_pose_detections = pose_detector.predict_on_image(img2)
pose_detections = denormalize_detections(normalized_pose_detections, scale,
pad)
xc, yc, scale, theta = pose_detector.detection2roi(pose_detections)
img, affine, box = pose_regressor.extract_roi(frame, xc, yc, theta, scale)
flags, normalized_landmarks, mask = pose_regressor(img.to(gpu))
landmarks = pose_regressor.denormalize_landmarks(normalized_landmarks,
affine)
draw_detections(frame, pose_detections)
draw_roi(frame, box)
for i in range(len(flags)):
landmark, flag = landmarks[i], flags[i]
if flag > .5:
draw_landmarks(frame, landmark, POSE_CONNECTIONS, size=2)
cv2.imshow(WINDOW, frame[:, :, ::-1])
# cv2.imwrite('sample/%04d.jpg'%frame_ct, frame[:,:,::-1])
hasFrame, frame = capture.read()
key = cv2.waitKey(1)
if key == 27:
break
capture.release()
cv2.destroyAllWindows()
img1 = torch.tensor(img1, device=gpu).byte()
img1 = img1.unsqueeze(0)
normalized_landmarks2, flags2 = hand_regressor(img1)
#for i in range(len(flags)):
# landmark, flag = landmarks[i], flags[i]
# if flag>.5:
# draw_landmarks(frame, landmark[:,:2], FACE_CONNECTIONS, size=1)
for i in range(len(flags2)):
landmark, flag = normalized_landmarks2[i], flags2[i]
#landmark, flag = landmarks2[i], flags2[i]
#if flag>.01:
if True:
draw_landmarks(frame, landmark[:,:2], HAND_CONNECTIONS, size=2)
print(flag)
#draw_landmarks(frame, landmark[:,:2] * 200, HAND_CONNECTIONS, size=2)
#draw_roi(frame, box)
#draw_roi(frame, box2)
#draw_detections(frame, face_detections)
#draw_detections(frame, palm_detections)
cv2.imshow(WINDOW, frame)
# cv2.imwrite('sample/%04d.jpg'%frame_ct, frame[:,:,::-1])
hasFrame, frame = capture.read()
key = cv2.waitKey(1)
if key == 27:
break
if imgs:
imgs = torch.stack(
imgs) #.permute(0,3,1,2).float() / 255. #/ 127.5 - 1.0
affines = torch.stack(affines)
else:
imgs = torch.zeros((0, 3, res, res)).to(gpu)
affines = torch.zeros((0, 2, 3)).to(gpu)
#img, affine2, box2 = hand_regressor.extract_roi(img1, xc, yc, theta, scale)
flags2, handed2, normalized_landmarks2 = hand_regressor(imgs)
landmarks2 = hand_regressor.denormalize_landmarks(normalized_landmarks2,
affines)
for i in range(len(flags2)):
landmark, flag = landmarks2[i], flags2[i]
if flag > .5:
draw_landmarks(img1, landmark[:, :2], HAND_CONNECTIONS, size=2)
draw_roi(img1, points)
cv2.imshow(WINDOW, img1)
hasFrame, frame = capture.read()
key = cv2.waitKey(1)
if key == 27:
break
capture.release()
cv2.destroyAllWindows()
input_name = ort_session.get_inputs()[0].name
while hasFrame:
frame_ct += 1
img1, img2, scale, pad = resize_pad(frame)
img3 = cv2.resize(img1, (192, 192))
img_in = np.expand_dims(img3, axis=0).astype(np.uint8)
ort_inputs = {input_name: img_in}
ort_outs = ort_session.run(None, ort_inputs)
frame = cv2.rectangle(img3, (30, 30), (162, 162), (255, 255, 255), -1)
img3 = cv2.resize(img3, (960, 960))
for i in range(len(ort_outs[0])):
landmark, flag = ort_outs[0][i], ort_outs[1][i]
if flag > .5:
draw_landmarks(img3, landmark[:, :2] * 5, FACE_CONNECTIONS, size=1)
cv2.imshow(WINDOW, img3)
hasFrame, frame = capture.read()
key = cv2.waitKey(1)
if key == 27:
break
capture.release()
cv2.destroyAllWindows()