Randaugment: Practical automated data augmentation with a reduced search space

RandAugment: практическое автоматизированное расширение данных с сокращённым пространством поиска
Quoc Viet Le, Barret Zoph, Jonathon Shlens, Ekin D. Cubuk
2020-06-01

RandAugmentautomated data augmentationimage classificationobject detectionreduced search space
Recent work on automated augmentation strategies has led to state-of-the-art results in image classification and object detection. An obstacle to a large-scale adoption of these methods is that they require a separate and expensive search phase. A common way to overcome the expense of the search phase was to use a smaller proxy task. However, it was not clear if the optimized hyperparameters found on the proxy task are also optimal for the actual task. In this work, we rethink the process of designing automated augmentation strategies. We find that while previous work required a search for both magnitude and probability of each operation independently, it is sufficient to only search for a single distortion magnitude that jointly controls all operations. We hence propose a simplified search space that vastly reduces the computational expense of automated augmentation, and permits the removal of a separate proxy task. Despite the simplifications, our method achieves equal or better performance over previous automated augmentation strategies on on CIFAR-10/100, SVHN, ImageNet and COCO datasets. EfficientNet-B7, we achieve 85.0% accuracy, a 1.0% increase over baseline augmentation, a 0.6% improvement over AutoAugment on the ImageNet dataset. With EfficientNet-B8, we achieve 85.4% accuracy on ImageNet, which matches a previous result that used 3.5B extra images. On object detection, the same method as classification leads to 1.0-1.3% improvement over baseline augmentation. Code will be made available online.
1
RandAugment matches or outperforms previous automated augmentation strategies on CIFAR-10/100, SVHN, ImageNet, and COCO.
2
RandAugment reduces automated augmentation search to a single distortion magnitude jointly controlling all operations, eliminating independent magnitude and probability searches.
3
The simplified search space substantially lowers computational cost and removes the need for a separate proxy task.
4
With EfficientNet-B7, RandAugment achieves 85.0% ImageNet accuracy, improving 1.0 percentage point over baseline augmentation and 0.6 points over AutoAugment.
5
With EfficientNet-B8, RandAugment reaches 85.4% ImageNet accuracy, matching a result that used 3.5 billion additional images; object detection improves 1.0–1.3 percentage points over baseline augmentation.

automated image augmentation strategies for image classification and object detection

the effectiveness and computational efficiency of a reduced search space using a single joint distortion magnitude instead of independently searching operation magnitudes and probabilities

Publication Details
Publication Date
2020-06-01
Journal
Publisher
ISSN
Cited by
5030
Access Type
Author Information
Authors
Quoc Viet Le
Barret Zoph
Jonathon Shlens
Ekin D. Cubuk
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%