Test-Time Data Augmentation
Tutorial on how to properly implement test-time image data augmentation in a production environment with limited computational resources.
data-augmentation keras production tensorflow deep-learning code notebook article tutorial

Proper test-time data augmentation in production usually does not apply the same augmentation pipeline as during training. In production, the computational budget is usually limited and we only have time to run inference on a few augmented images.

Therefore, the random data augmentation that works well during model training is not ideal. It is important to get the highest accuracy boost with only a few augmented samples. In practice this means that the best test-time data augmentation has to be found using experimentation

Run the Colab notebook to experiment for yourself and learn how to find the best augmentation parameters.

Don't forget to tag @dufourpascal in your comment, otherwise they may not be notified.

Authors original post
PhD in Biomedical Engineering. Computer scientist with a focus on computer vision and bringing deep learning to production. I blog over at stepup.ai
Share this project
Similar projects
Data augmentation recipes in tf.keras image-based models
Learn about different ways of doing data augmentation when training an image classifier in tf.keras.
Albumentations
Fast image augmentation library and easy to use wrapper around other libraries.
Automatic Data Augmentation for Generalization in Deep RL
We compare three approaches for automatically finding an appropriate augmentation combined with two novel regularization terms for the policy and value ...
Multi-target in Albumentations
Many images, many masks, bounding boxes, and key points. How to transform them in sync?