Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

[Bug] Mismatched input and output shapes in Augmentations

See original GitHub issue

🐛 Bug

To Reproduce

Steps to reproduce the behavior:

Simply applying ColorJitter to an RGB image would produce the batch version. First discussion was here.

>>> import torch
>>> from kornia.augmentation import ColorJitter
>>> x = T.rand(3, 4, 4)
>>> cj = ColorJitter(0.2, 0.3, 0.2, 0.3)
>>> cj(x).shape
torch.Size([1, 3, 4, 4])

Expected behavior

Output should have the same shape as input.

Additional context

Maybe it is the originally intended behavior, but I think it is too rigid and deeply integrated into training deep NN, while the goal of Kornia is a differentiable OpenCV in GPU. Or am I missing anything here?

Edit 2: Added link to the original discussion.

Issue Analytics

State:
Created 3 years ago
Comments:6 (3 by maintainers)

Top GitHub Comments

2reactions

justanhduccommented, Oct 20, 2020

Ok I will try to work on it asap.

0reactions

shijianjiancommented, Nov 22, 2020

Cloased by #731

Top Results From Across the Web

Errors when using imgaug's augmentation. #2018 - GitHub

Hello, I'm getting this error each time I'm trying to use augmentation. SuspiciousSingleImageShapeWarning: You provided a numpy array of shape ( ...

Data Augmentation Layer in Keras Sequential Model

I tried specifying the input shape in the augmented layer as well. When I take out the data_augmentation layer from the model it...

Change input shape dimensions for fine-tuning with Keras

In this tutorial, you will learn how to change the input shape tensor dimensions for fine-tuning using Keras. After going through this guide ......

How to Configure Image Data Augmentation in Keras

The brightness of the image can be augmented by either randomly darkening images, brightening images, or both. The intent is to allow a...

Working with preprocessing layers - Keras

With Keras preprocessing layers, you can build and export models that are truly end-to-end: models that accept raw images or raw structured data ......