def convert_video_mjpeg(cls, file_path: str, delete_mjpeg=False):
extension = os.path.splitext(file_path)[1]
if extension != '.mjpeg':
raise Exception('file_path must have extension .mjpeg')
else:
output_video = file_path.replace('.mjpeg', '.mjpegx')
print('output_video: ' + output_video)
# Converting first with _1 to avoid confusions
file_path_temp = output_video + '_1'
print('Initializing instance')
open_pose = ClassOpenPose()
images = ClassMjpegReader.process_video(file_path)
print('Len images: ' + str(len(images)))
with open(file_path_temp, 'wb') as newFile:
counter = 0
for elem in images:
image = elem[0]
ticks = elem[1]
image_np = np.frombuffer(image, dtype=np.uint8)
image_cv = cv2.imdecode(image_np, cv2.IMREAD_ANYCOLOR)
arr = open_pose.recognize_image(
image_cv) # type: np.ndarray
json_dict = {'vectors': arr.tolist()}
ClassUtils.write_in_file(newFile, ticks, image, json_dict)
counter += 1
if counter % 10 == 0:
print('Counter: ' + str(counter))
# Deleting mjpegx file if exists and rename
if os.path.exists(output_video):
os.remove(output_video)
# Naming new
os.rename(file_path_temp, output_video)
if delete_mjpeg:
os.remove(file_path)
def save_video_from_list_frames(cls, file_path: str, list_frames):
extension = os.path.splitext(file_path)[1]
if extension != '.mjpegx':
raise Exception('file_path must have extension .mjpegx')
output_video = file_path + '_1'
with open(output_video, 'wb') as newFile:
counter = 0
for elem in list_frames:
# Avoid previous conversion -> only reading
image = elem[0]
ticks = elem[1]
json_dict = elem[2]
# Write in new file
ClassUtils.write_in_file(newFile, ticks, image, json_dict)
# Delete old
os.remove(file_path)
# Naming new
os.rename(output_video, file_path)
def convert_video_mjpegx(cls,
file_path: str,
cam_number: str,
instance_nn_pose: ClassNN = None):
extension = os.path.splitext(file_path)[1]
if ClassUtils.cam_calib_exists(cam_number):
calib_params = ClassUtils.load_cam_calib_params(cam_number)
else:
print('Warning: Calib params not found for camera {0}'.format(
cam_number))
calib_params = None
if extension != '.mjpegx':
raise Exception('file_path must have extension .mjpegx')
print('Reading image')
output_video = file_path + '_1'
images = ClassMjpegReader.process_video_mjpegx(file_path)
print('Len images: ' + str(len(images)))
with open(output_video, 'wb') as newFile:
counter = 0
for elem in images:
# Avoid previous conversion -> only reading
image = elem[0]
ticks = elem[1]
json_dict = elem[2]
vectors = json_dict['vectors']
# Avoid over processing - Only process images with poses
image_cv = None
if len(vectors) != 0:
image_np = np.frombuffer(image, dtype=np.uint8)
image_cv = cv2.imdecode(image_np, cv2.IMREAD_ANYCOLOR)
params = list()
for vector in vectors:
# Check if vector is bodypart 25
# Avoid confusions with COCO processing
if len(vector) != 25:
raise Exception(
'Vector is not bodypart 25 - File: {0} Len Vectors: {1}'
.format(file_path, len(vector)))
params.append(
ClassDescriptors.get_person_descriptors(
vector,
cls.min_percent,
cam_number=cam_number,
image=image_cv,
calib_params=calib_params,
decode_img=False,
instance_nn_pose=instance_nn_pose))
# Save object globally
# CamNumber
new_json_dict = {
'vectors': vectors,
'params': params,
'camNumber': cam_number
}
# Write in new file - New json dict
ClassUtils.write_in_file(newFile, ticks, image, new_json_dict)
counter += 1
if counter % 100 == 0:
print('Counter: ' + str(counter))
if calib_params is None:
print('Warning: Calib params not found for camera {0}'.
format(cam_number))
# Delete old
os.remove(file_path)
# Naming new
os.rename(output_video, file_path)
def convert_video_mjpegx_reid(cls,
file_path: str,
list_people_reid=None,
save_img=False,
option=None):
extension = os.path.splitext(file_path)[1]
if extension != '.mjpegx':
raise Exception('file_path must have extension .mjpegx')
print('Reading image')
output_video = file_path + '_1'
images = ClassMjpegReader.process_video_mjpegx(file_path)
print('Len images: ' + str(len(images)))
with open(output_video, 'wb') as newFile:
counter = 0
for elem in images:
# Avoid previous conversion -> only reading
image = elem[0]
ticks = elem[1]
json_dict = elem[2]
params = json_dict['params']
# Update elements
for param in params:
found = False
for person_reid in list_people_reid:
for item in person_reid.list_poses:
if item['poseGuid'] == param['poseGuid']:
param['personGuid'] = person_reid.person_guid
found = True
break
if found:
# Saving image with pose identified
if save_img:
cnn_folder = ClassUtils.cnn_folder
if option is not None:
cnn_folder = os.path.join(
cnn_folder, option)
path_img = os.path.join(
cnn_folder, param['personGuid'])
if not os.path.exists(path_img):
os.makedirs(path_img)
filename = os.path.join(
path_img, '{0}.jpg'.format(ticks))
image_arr = np.frombuffer(image,
dtype=np.uint8)
image_cv = cv2.imdecode(
image_arr, cv2.IMREAD_ANYCOLOR)
ClassDescriptors.draw_pose(
image_cv, param['vectors'],
cls.min_percent, param['keyPose'])
cv2.imwrite(filename, image_cv)
break
if not found:
param['personGuid'] = ''
# Saving data in json dict
ClassUtils.write_in_file(newFile, ticks, image, json_dict)
counter += 1
if counter % 100 == 0:
print('Counter: ' + str(counter))
# Delete old
os.remove(file_path)
# Naming new
os.rename(output_video, file_path)