Visualise images before they enter the network

[benedetta.delfino]

Hello,

I am training a model with TLT, is it possible to visualise the images just before they enter the network for training/ inference? It would help with debugging/ checking for data augmentation.

Similarly, is it possible to visualise images in deepstream before inference?

Thanks

[Morganh]

@benedetta.delfino

images just before they enter the network for training/ inference

@benedetta.delfino

Similarly, is it possible to visualise images in deepstream before inference?

For your above questions, there should be the original images from your training dataset, right? You can visualize them.

But for the dataloader online augmentation, there is not visualization api provided by TAO.