training/built-in-frameworks/fastai_oxford_pets/source/pets.py

import argparse
import glob
import io
import logging
import os
import time

import requests
import sagemaker_containers
from fastai.vision import *

logger = logging.getLogger(__name__)
logger.setLevel(logging.DEBUG)

# set the constants for the content types
JSON_CONTENT_TYPE = "application/json"
JPEG_CONTENT_TYPE = "image/jpeg"


def _train(args):
    device = "cuda" if torch.cuda.is_available() else "cpu"
    logger.info("Device Type: {}".format(device))

    logger.info("Loading Pets dataset")
    print(f"Batch size: {args.batch_size}")
    path = Path(args.data_dir)
    print(f"Data path is: {path}")
    path_anno = path / "annotations"
    path_img = path / "images"
    fnames = get_image_files(path_img)

    # get the pattern to select the training/validation data
    np.random.seed(2)
    pat = re.compile(r"/([^/]+)_\d+.jpg$")
    print("Creating DataBunch object")
    data = ImageDataBunch.from_name_re(
        path_img, fnames, pat, ds_tfms=get_transforms(), size=args.image_size, bs=args.batch_size
    ).normalize(imagenet_stats)

    # create the CNN model
    print("Create CNN model from model zoo")
    print(f"Model architecture is {args.model_arch}")
    arch = getattr(models, args.model_arch)
    print("Creating pretrained conv net")
    learn = create_cnn(data, arch, metrics=error_rate)
    print("Fit for 4 cycles")
    learn.fit_one_cycle(4)
    learn.unfreeze()
    print("Unfreeze and fit for another 2 cycles")
    learn.fit_one_cycle(2, max_lr=slice(1e-6, 1e-4))
    print("Finished Training")

    logger.info("Saving the model.")
    model_path = Path(args.model_dir)
    print(f"Export data object")
    data.export(model_path / "export.pkl")
    # create empty models dir
    os.mkdir(model_path / "models")
    print(f"Saving model weights")
    return learn.save(model_path / f"{args.model_arch}")


def model_fn(model_dir):
    logger.info("model_fn")
    path = Path(model_dir)
    print("Creating DataBunch object")
    empty_data = ImageDataBunch.load_empty(path)
    arch_name = os.path.splitext(os.path.split(glob.glob(f"{model_dir}/resnet*.pth")[0])[1])[0]
    print(f"Model architecture is: {arch_name}")
    arch = getattr(models, arch_name)
    learn = create_cnn(empty_data, arch, pretrained=False).load(path / f"{arch_name}")
    return learn


# Deserialize the Invoke request body into an object we can perform prediction on
def input_fn(request_body, content_type=JPEG_CONTENT_TYPE):
    logger.info("Deserializing the input data.")
    # process an image uploaded to the endpoint
    if content_type == JPEG_CONTENT_TYPE:
        img = open_image(io.BytesIO(request_body))
        return img
    # process a URL submitted to the endpoint
    if content_type == JSON_CONTENT_TYPE:
        img_request = requests.get(request_body["url"], stream=True)
        img = open_image(io.BytesIO(img_request.content))
        return img
    raise Exception("Requested unsupported ContentType in content_type: {}".format(content_type))


# Perform prediction on the deserialized object, with the loaded model
def predict_fn(input_object, model):
    logger.info("Calling model")
    start_time = time.time()
    predict_class, predict_idx, predict_values = model.predict(input_object)
    print("--- Inference time: %s seconds ---" % (time.time() - start_time))
    print(f"Predicted class is {str(predict_class)}")
    print(f"Predict confidence score is {predict_values[predict_idx.item()].item()}")
    response = {}
    response["class"] = str(predict_class)
    response["confidence"] = predict_values[predict_idx.item()].item()
    return response


# Serialize the prediction result into the desired response content type
def output_fn(prediction, accept=JSON_CONTENT_TYPE):
    logger.info("Serializing the generated output.")
    if accept == JSON_CONTENT_TYPE:
        output = json.dumps(prediction)
        return output, accept
    raise Exception("Requested unsupported ContentType in Accept: {}".format(accept))


if __name__ == "__main__":
    parser = argparse.ArgumentParser()

    parser.add_argument(
        "--workers",
        type=int,
        default=2,
        metavar="W",
        help="number of data loading workers (default: 2)",
    )
    parser.add_argument(
        "--epochs",
        type=int,
        default=2,
        metavar="E",
        help="number of total epochs to run (default: 2)",
    )
    parser.add_argument(
        "--batch_size", type=int, default=64, metavar="BS", help="batch size (default: 4)"
    )
    parser.add_argument(
        "--lr",
        type=float,
        default=0.001,
        metavar="LR",
        help="initial learning rate (default: 0.001)",
    )
    parser.add_argument(
        "--momentum", type=float, default=0.9, metavar="M", help="momentum (default: 0.9)"
    )
    parser.add_argument(
        "--dist_backend", type=str, default="gloo", help="distributed backend (default: gloo)"
    )

    # fast.ai specific parameters
    parser.add_argument(
        "--image-size", type=int, default=224, metavar="IS", help="image size (default: 224)"
    )
    parser.add_argument(
        "--model-arch",
        type=str,
        default="resnet34",
        metavar="MA",
        help="model arch (default: resnet34)",
    )

    env = sagemaker_containers.training_env()
    parser.add_argument("--hosts", type=list, default=env.hosts)
    parser.add_argument("--current-host", type=str, default=env.current_host)
    parser.add_argument("--model-dir", type=str, default=env.model_dir)
    parser.add_argument("--data-dir", type=str, default=env.channel_input_dirs.get("training"))
    parser.add_argument("--num-gpus", type=int, default=env.num_gpus)

    _train(parser.parse_args())