How to use openpifpaf - 5 common examples

def cli():
    parser = argparse.ArgumentParser(
        description=__doc__,
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )
    openpifpaf.decoder.cli(parser, force_complete_pose=False,
                           instance_threshold=0.1, seed_threshold=0.5)
    openpifpaf.network.nets.cli(parser)
    parser.add_argument('--disable-cuda', action='store_true',
                        help='disable CUDA')
    parser.add_argument('--resolution', default=0.4, type=float,
                        help=('Resolution prescale factor from 640x480. '
                              'Will be rounded to multiples of 16.'))
    parser.add_argument('--write-static-page', default=None,
                        help='directory in which to create a static version of this page')
    parser.add_argument('--debug', default=False, action='store_true',
                        help='debug messages and autoreload')
    parser.add_argument('--google-analytics',
                        help='provide a google analytics id to inject analytics code')

    parser.add_argument('--host', dest='host',
                        default=os.environ.get('HOST', '127.0.0.1'),

def cli():
    parser = argparse.ArgumentParser(
        description=__doc__,
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
    )
    openpifpaf.decoder.cli(parser, force_complete_pose=False,
                           instance_threshold=0.1, seed_threshold=0.5)
    openpifpaf.network.nets.cli(parser)
    parser.add_argument('--disable-cuda', action='store_true',
                        help='disable CUDA')
    parser.add_argument('--resolution', default=0.4, type=float,
                        help=('Resolution prescale factor from 640x480. '
                              'Will be rounded to multiples of 16.'))
    parser.add_argument('--write-static-page', default=None,
                        help='directory in which to create a static version of this page')
    parser.add_argument('--debug', default=False, action='store_true',
                        help='debug messages and autoreload')
    parser.add_argument('--google-analytics',
                        help='provide a google analytics id to inject analytics code')

    parser.add_argument('--host', dest='host',
                        default=os.environ.get('HOST', '127.0.0.1'),
                        help='host address for webserver, use 0.0.0.0 for global access')
    parser.add_argument('--port', dest='port',

def __init__(self, width_height, args):
        self.width_height = width_height

        # load model
        self.model, _ = openpifpaf.network.nets.factory_from_args(args)
        self.model = self.model.to(args.device)
        self.processor = openpifpaf.decoder.factory_from_args(args, self.model)
        self.device = args.device

def __init__(self, width_height, args):
        self.width_height = width_height

        # load model
        self.model, _ = openpifpaf.network.nets.factory_from_args(args)
        self.model = self.model.to(args.device)
        self.processor = openpifpaf.decoder.factory_from_args(args, self.model)
        self.device = args.device

def single_image(self, b64image):
        imgstr = re.search(r'base64,(.*)', b64image).group(1)
        image_bytes = io.BytesIO(base64.b64decode(imgstr))
        im = PIL.Image.open(image_bytes).convert('RGB')

        target_wh = self.width_height
        if (im.size[0] > im.size[1]) != (target_wh[0] > target_wh[1]):
            target_wh = (target_wh[1], target_wh[0])
        if im.size[0] != target_wh[0] or im.size[1] != target_wh[1]:
            print('!!! have to resize image to', target_wh, ' from ', im.size)
            im = im.resize(target_wh, PIL.Image.BICUBIC)
        width_height = im.size

        start = time.time()
        preprocess = openpifpaf.transforms.EVAL_TRANSFORM
        processed_image_cpu, _, __ = preprocess(im, [], None)
        processed_image = processed_image_cpu.contiguous().to(self.device, non_blocking=True)
        print('preprocessing time', time.time() - start)

        all_fields = self.processor.fields(torch.unsqueeze(processed_image.float(), 0))[0]
        keypoint_sets, scores = self.processor.keypoint_sets(all_fields)

        # normalize scale
        keypoint_sets[:, :, 0] /= processed_image_cpu.shape[2]
        keypoint_sets[:, :, 1] /= processed_image_cpu.shape[1]

        return keypoint_sets, scores, width_height