Python comm.is_main_process方法代码示例

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def _accumulate_predictions_from_multiple_gpus(predictions_per_gpu):
    all_predictions = scatter_gather(predictions_per_gpu)
    if not is_main_process():
        return
    # merge the list of dicts
    predictions = {}
    for p in all_predictions:
        predictions.update(p)
    # convert a dict where the key is the index in a list
    image_ids = list(sorted(predictions.keys()))
    if len(image_ids) != image_ids[-1] + 1:
        logger = logging.getLogger("maskrcnn_benchmark.eval_IR")
        logger.warning(
            "Number of images that were gathered from multiple processes is not "
            "a contiguous set. Some images might be missing from the evaluation"
        )

    # convert to a list
    predictions = [predictions[i] for i in image_ids]
    return predictions 

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def test_and_exchange_map(tester, model, distributed):
    results = tester(model=model, distributed=distributed)

    # main process only
    if is_main_process():
        # Note: one indirection due to possibility of multiple test datasets, we only care about the first
        #       tester returns (parsed results, raw results). In our case, don't care about the latter
        map_results, raw_results = results[0]
        bbox_map = map_results.results["bbox"]['AP']
        segm_map = map_results.results["segm"]['AP']
    else:
        bbox_map = 0.
        segm_map = 0.

    if distributed:
        map_tensor = torch.tensor([bbox_map, segm_map], dtype=torch.float32, device=torch.device("cuda"))
        torch.distributed.broadcast(map_tensor, 0)
        bbox_map = map_tensor[0].item()
        segm_map = map_tensor[1].item()

    return bbox_map, segm_map 

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def print_mlperf(key, value=None):
    if is_main_process():
        maskrcnn_print(key=key, value=value) 

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def cache_url(url, model_dir=None, progress=True):
    r"""Loads the Torch serialized object at the given URL.
    If the object is already present in `model_dir`, it's deserialized and
    returned. The filename part of the URL should follow the naming convention
    ``filename-<sha256>.ext`` where ``<sha256>`` is the first eight or more
    digits of the SHA256 hash of the contents of the file. The hash is used to
    ensure unique names and to verify the contents of the file.
    The default value of `model_dir` is ``$TORCH_HOME/models`` where
    ``$TORCH_HOME`` defaults to ``~/.torch``. The default directory can be
    overridden with the ``$TORCH_MODEL_ZOO`` environment variable.
    Args:
        url (string): URL of the object to download
        model_dir (string, optional): directory in which to save the object
        progress (bool, optional): whether or not to display a progress bar to stderr
    Example:
        >>> cached_file = maskrcnn_benchmark.utils.model_zoo.cache_url('https://s3.amazonaws.com/pytorch/models/resnet18-5c106cde.pth')
    """
    if model_dir is None:
        torch_home = os.path.expanduser(os.getenv('TORCH_HOME', '~/.torch'))
        model_dir = os.getenv('TORCH_MODEL_ZOO', os.path.join(torch_home, 'models'))
    if not os.path.exists(model_dir):
        os.makedirs(model_dir)
    parts = urlparse(url)
    filename = os.path.basename(parts.path)
    if filename == "model_final.pkl":
        # workaround as pre-trained Caffe2 models from Detectron have all the same filename
        # so make the full path the filename by replacing / with _
        filename = parts.path.replace("/", "_")
    cached_file = os.path.join(model_dir, filename)
    if not os.path.exists(cached_file) and is_main_process():
        sys.stderr.write('Downloading: "{}" to {}\n'.format(url, cached_file))
        hash_prefix = HASH_REGEX.search(filename)
        if hash_prefix is not None:
            hash_prefix = hash_prefix.group(1)
            # workaround: Caffe2 models don't have a hash, but follow the R-50 convention,
            # which matches the hash PyTorch uses. So we skip the hash matching
            # if the hash_prefix is less than 6 characters
            if len(hash_prefix) < 6:
                hash_prefix = None
        _download_url_to_file(url, cached_file, hash_prefix, progress=progress)
    synchronize()
    return cached_file 

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def cache_url(url, model_dir=None, progress=True):
    r"""Loads the Torch serialized object at the given URL.
    If the object is already present in `model_dir`, it's deserialized and
    returned. The filename part of the URL should follow the naming convention
    ``filename-<sha256>.ext`` where ``<sha256>`` is the first eight or more
    digits of the SHA256 hash of the contents of the file. The hash is used to
    ensure unique names and to verify the contents of the file.
    The default value of `model_dir` is ``$TORCH_HOME/models`` where
    ``$TORCH_HOME`` defaults to ``~/.torch``. The default directory can be
    overridden with the ``$TORCH_MODEL_ZOO`` environment variable.
    Args:
        url (string): URL of the object to download
        model_dir (string, optional): directory in which to save the object
        progress (bool, optional): whether or not to display a progress bar to stderr
    Example:
        >>> cached_file = maskrcnn_benchmark.utils.model_zoo.cache_url('https://s3.amazonaws.com/pytorch/models/resnet18-5c106cde.pth')
    """
    if model_dir is None:
        torch_home = os.path.expanduser(os.getenv("TORCH_HOME", "~/.torch"))
        model_dir = os.getenv("TORCH_MODEL_ZOO", os.path.join(torch_home, "models"))
    if not os.path.exists(model_dir):
        os.makedirs(model_dir)
    parts = urlparse(url)
    filename = os.path.basename(parts.path)
    if filename == "model_final.pkl":
        # workaround as pre-trained Caffe2 models from Detectron have all the same filename
        # so make the full path the filename by replacing / with _
        filename = parts.path.replace("/", "_")
    cached_file = os.path.join(model_dir, filename)
    if not os.path.exists(cached_file) and is_main_process():
        sys.stderr.write('Downloading: "{}" to {}\n'.format(url, cached_file))
        hash_prefix = HASH_REGEX.search(filename)
        if hash_prefix is not None:
            hash_prefix = hash_prefix.group(1)
            # workaround: Caffe2 models don't have a hash, but follow the R-50 convention,
            # which matches the hash PyTorch uses. So we skip the hash matching
            # if the hash_prefix is less than 6 characters
            if len(hash_prefix) < 6:
                hash_prefix = None
        _download_url_to_file(url, cached_file, hash_prefix, progress=progress)
    synchronize()
    return cached_file 

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def inference(
    model,
    rngs,
    data_loader,
    iou_types=("bbox",),
    box_only=False,
    device="cuda",
    expected_results=(),
    expected_results_sigma_tol=4,
    output_folder=None,
):

    # convert to a torch.device for efficiency
    device = torch.device(device)
    num_devices = (
        torch.distributed.get_world_size()
        if torch.distributed.is_initialized()
        else 1
    )
    dataset = data_loader.dataset
    predictions = compute_on_dataset(model, rngs, data_loader, device)
    # wait for all processes to complete before measuring the time
    synchronize()

    predictions = _accumulate_predictions_from_multiple_gpus(predictions)
    if not is_main_process():
        return

    coco_results = {}
    if "bbox" in iou_types:
        coco_results["bbox"] = prepare_for_coco_detection(predictions, dataset)

    results = COCOResults(*iou_types)
    uuid1 = str(uuid.uuid1())

    for iou_type in iou_types:
        with tempfile.NamedTemporaryFile() as f:
            file_path = f.name
            if output_folder:
                file_path = os.path.join(
                    output_folder, uuid1 + iou_type + ".json")
            res = evaluate_predictions_on_coco(
                dataset.coco, coco_results[iou_type], file_path, iou_type
            )
            results.update(res)

        if os.path.isfile(file_path):
            os.remove(file_path)

    return results 

# 需要导入模块: from maskrcnn_benchmark.utils import comm [as 别名]
# 或者: from maskrcnn_benchmark.utils.comm import is_main_process [as 别名]
def inference(
    model,
    data_loader,
    iou_types=("bbox",),
    box_only=False,
    device="cuda",
    expected_results=(),
    expected_results_sigma_tol=4,
    without_nms=False,
    output_folder=None,
):

    # convert to a torch.device for efficiency
    device = torch.device(device)
    num_devices = (
        torch.distributed.get_world_size()
        if torch.distributed.is_initialized()
        else 1
    )
    logger = logging.getLogger("maskrcnn_benchmark.eval_IR")
    dataset = data_loader.dataset
    logger.info("Start evaluation on {} images".format(len(dataset)))
    start_time = time.time()
    with_overlaps, without_overlaps = compute_on_dataset(model, data_loader, device)
    # wait for all processes to complete before measuring the time
    synchronize()
    total_time = time.time() - start_time
    total_time_str = str(datetime.timedelta(seconds=total_time))
    logger.info(
        "Total inference time: {} ({} s / img per device, on {} devices)".format(
            total_time_str, total_time * num_devices / len(dataset), num_devices
        )
    )

    with_overlaps = _accumulate_predictions_from_multiple_gpus(with_overlaps)
    without_overlaps = _accumulate_predictions_from_multiple_gpus(without_overlaps)
    if not is_main_process():
        return

    logger.info("Evaluating IoU average Recall (IR)")
    results = evaluate_iou_average_recall(with_overlaps, without_overlaps, output_folder)
    logger.info(results)