A while ago I inquired about support for grayscale images in TLT 2.0. At that time grayscale was not supported. However, in TAO there seems to be a support for grayscale again, specifically in the Faster RCNN config spec. In order for this to work, there would also need to be a grayscale backbone available on the ngc for use as a pretrained model, but I was not able to find any such backbone in the catalog. I was wondering if grayscale is now supported in TAO and, if so, is there a pretrained model for supporting grayscale or does the model need to be trained from randomized weights.
TAO can support training with grayscale images.
But in ngc, we did not provide grayscale pretrained models.
Currently, end user can train from scratch.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.