add color jitter augmentation #680
Conversation
facebook-github-bot
added
the
CLA Signed
| # Image ColorJitter | ||
| _C.INPUT.BRIGHTNETSS = 0.0 | ||
| _C.INPUT.CONTRAST = 0.0 | ||
| _C.INPUT.SITURATION = 0.0 |
There was a problem hiding this comment.
typo: it should be SATURATION
| @@ -54,6 +54,12 @@ | |||
| # Convert image to BGR format (for Caffe2 models), in range 0-255 | |||
| _C.INPUT.TO_BGR255 = True | |||
|
|
|||
| # Image ColorJitter | |||
| _C.INPUT.BRIGHTNETSS = 0.0 | |||
There was a problem hiding this comment.
Hi,
I have correct the spelling, please tell me if there needs further modifications.
As for the performance, I actually observed no obvious improvements on our own datasets with brightness=0.4, contrast=0.4, saturation=0.4, hue=0.3. The random seed is not manually assigned and the performance is on par with the non-color-aug version.
There was a problem hiding this comment.
By the way, do you think we should add a random_seed field in the configuration file for convenience of assigning the seed for random operations?
There was a problem hiding this comment.
Having a seed would be useful I think, even though not everything would be deterministic.
|
Hi, Is there further modification requirements, or is this pull request not really necessary ? |
* add color jitter augmentation * fix spelling (cherry picked from commit 862347d)
* add color jitter augmentation * fix spelling
Inspired by this paper, maybe the augmentation of adjusting color, brightness, contrast and saturation will make the model get trained more stably.