Augmentations during training

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

Hi! I use sample configs, which provide augmentation_config part. If I don’t use tlt-augment offline (that is - BEFORE the training) does that mean that I run the training without any augmentations?

2

No, there is still some augmentations.
For example, Integrating TAO Models into DeepStream — TAO Toolkit 3.22.05 documentation

3

Hm, I see. Then what’s the profit of using your augment tool, if we can use the same augmentations during training?

4

See more details in Integrating TAO Models into DeepStream — TAO Toolkit 3.22.05 documentation

5

Would it possible to use tlt-augment to resize image to the same size from VOC or COCO?

6

Actually, it can. Please remove the spatial and color part inside the tlt-augment spec.