My train dataset contains images with labels mentioning whether the images are rotated left or rotated right or upright or upside down. My test set has no labels, I need to rotate all my test images upright. How do I go about this?
so you have to predict which way the test image is rotated from the trained model and then actually rotate it?
@joedockrill yes so I am predicting the labels on test images right now. So I would get what image is upside down or rotated left or rotated right, etc.
How do I rotate it once I have the predictions. As in, if the image is rotated left, I need to rotate it right. If it is upside down, I need to rotate it to make it upright.
Okay I understood, thank you !