def test_invalid_image_dataset():
ImagePredictor = Task
invalid_test = ag.download('https://autogluon.s3-us-west-2.amazonaws.com/miscs/test_autogluon_invalid_dataset.zip')
invalid_test = ag.unzip(invalid_test)
df = ImagePredictor.Dataset.from_csv(os.path.join(invalid_test, 'train.csv'), root=os.path.join(invalid_test, 'train_images'))
predictor = ImagePredictor(label="labels")
predictor.fit(df, df.copy(), time_limit=60)
def download_shopee(dataset, data_path):
if not os.path.exists(os.path.join(data_path, dataset + '.zip')):
filename = ag.download(
'https://autogluon.s3.amazonaws.com/datasets/shopee-iet.zip',
path='data/shopee-iet.zip')
ag.mkdir(filename[:-4])
ag.unzip(filename, root=filename[:-4])
else:
print(dataset + '.zip already exists.\n')
def load_data(directory_prefix, train_file, test_file, name, url=None):
if not os.path.exists(directory_prefix):
os.mkdir(directory_prefix)
directory = directory_prefix + name + "/"
train_file_path = directory + train_file
test_file_path = directory + test_file
if (not os.path.exists(train_file_path)) or (not os.path.exists(test_file_path)):
# fetch files from s3:
print("%s data not found locally, so fetching from %s" % (name, url))
zip_name = ag.download(url, directory_prefix)
ag.unzip(zip_name, directory_prefix)
os.remove(zip_name)
train_data = task.Dataset(file_path=train_file_path)
test_data = task.Dataset(file_path=test_file_path)
return train_data, test_data
def get_dataset(args):
# built-in dataset (voc)
if 'voc' in args.dataset_name:
logging.info('Please follow this instruction to download dataset: \
https://gluon-cv.mxnet.io/build/examples_datasets/pascal_voc.html#sphx-glr-build-examples-datasets-pascal-voc-py '
)
train_dataset = task.Dataset(name=args.dataset_name)
test_dataset = task.Dataset(name=args.dataset_name, Train=False)
return (train_dataset, test_dataset)
# custom datset.
if args.dataset_name in dataset_dict:
url, index_file_name_trainval, index_file_name_test, classes, \
= dataset_dict[args.dataset_name]
data_root = os.path.join(args.dataset_root, args.dataset_name)
if not args.no_redownload:
root = args.dataset_root
filename_zip = ag.download(url, path=root)
filename = ag.unzip(filename_zip, root=root)
data_root = os.path.join(root, filename)
else:
logging.info(
"This dataset is not in dataset_dict. It should be downloaded before running this script."
)
index_file_name_trainval = args.index_file_name_trainval
index_file_name_test = args.index_file_name_test
classes = args.classes
data_root = args.data_root
train_dataset = task.Dataset(data_root,
index_file_name=index_file_name_trainval,
classes=classes)
test_dataset = task.Dataset(data_root,
index_file_name=index_file_name_test,
classes=classes,
Train=False)
return (train_dataset, test_dataset)
def get_dataset(args):
# built-in dataset (voc)
if 'voc' in args.dataset_name:
logging.info('Please follow this instruction to download dataset: \
https://gluon-cv.mxnet.io/build/examples_datasets/pascal_voc.html#sphx-glr-build-examples-datasets-pascal-voc-py '
)
root = os.path.expanduser('~/.mxnet/datasets/voc/VOC2007')
train_dataset = ObjectDetector.Dataset.from_voc(root,
splits='trainval')
test_dataset = ObjectDetector.Dataset.from_voc(root, splits='test')
return (train_dataset, test_dataset)
# custom datset.
if args.dataset_name in dataset_dict:
url, index_file_name_trainval, index_file_name_test, classes, \
= dataset_dict[args.dataset_name]
data_root = os.path.join(args.dataset_root, args.dataset_name)
if not args.no_redownload:
root = args.dataset_root
filename_zip = ag.download(url, path=root)
filename = ag.unzip(filename_zip, root=root)
data_root = os.path.join(root, filename)
else:
logging.info(
"This dataset is not in dataset_dict. It should be downloaded before running this script."
)
index_file_name_trainval = args.index_file_name_trainval
index_file_name_test = args.index_file_name_test
data_root = args.data_root
train_dataset = ObjectDetector.Dataset.from_voc(
data_root, splits=index_file_name_trainval)
test_dataset = ObjectDetector.Dataset.from_voc(data_root,
splits=index_file_name_test)
return (train_dataset, test_dataset)
import autogluon.core as ag
from autogluon.vision.object_detection import ObjectDetection as task
import os
root = './'
filename_zip = ag.download('https://autogluon.s3.amazonaws.com/datasets/tiny_motorbike.zip',
path=root)
filename = ag.unzip(filename_zip, root=root)
data_root = os.path.join(root, filename)
dataset_train = task.Dataset(data_root, classes=('motorbike',))
time_limits = 5*60*60 # 5 hours
epochs = 30
detector = task.fit(dataset_train,
num_trials=2,
epochs=epochs,
lr=ag.Categorical(5e-4, 1e-4),
ngpus_per_trial=1,
time_limits=time_limits)
# Evaluation on test dataset
dataset_test = task.Dataset(data_root, index_file_name='test', classes=('motorbike',))
test_map = detector.evaluate(dataset_test)
print("mAP on test dataset: {}".format(test_map[1][1]))
# visualization
image = '000467.jpg'
image_path = os.path.join(data_root, 'JPEGImages', image)
print(image_path)
