MXbox: Simple, efficient and flexible vision toolbox for mxnet framework.

MXbox is a toolbox aiming to provide a general and simple interface for vision tasks. This project is greatly inspired by PyTorch and torchvision. Detailed copyright files are on the way. Improvements and suggestions are welcome.

Installation

pip install mxbox

Features

1. Define preprocess as a flow

transform = transforms.Compose([
    transforms.RandomSizedCrop(224),
    transforms.RandomHorizontalFlip(),
    transforms.mx.ToNdArray(),
    transforms.mx.Normalize(mean = [ 0.485, 0.456, 0.406 ],
                            std  = [ 0.229, 0.224, 0.225 ]),
])

PS: By default, mxbox uses PIL to read and transform images. But it also supports other backends like accimage and skimage.

More examples can be found in XXX.

2. Build DataLoader in several lines

feedin_shapes = {
    'batch_size': 8,
    'data': [mx.io.DataDesc(name='data', shape=(8, 3, 32, 32), layout='NCHW')],
    'label': [mx.io.DataDesc(name='softmax_label', shape=(8, 1), layout='N')]
}

dst = Dataset(root='../../data', transform=img_transform, label_transform=label_transform)
loader = DataLoader(dst, feedin_shapes, threads=8, shuffle=True)

Also, common datasets such as cifar10, cifar100, SVHN, MNIST are out-of-the-box. You can simply load them from mxbox.datasets.

3. Load popular model with pretrained weights

vgg = mxbox.models.vgg(num_classes=10, pretrained=True)
resnet = mxbox.models.resnet152(num_classes=10, pretrained=True)

Documentation

Under construction, coming soon.

TODO list

1. Efficient multi-thread reading (Prefetch wanted

2. Common Models preparation.

3. More friendly error logging.