TFDV - Unstructured Data

Can TFDV be used for unstructured data sets such as video, audio and signals? If not, what would be a good tool? For video for example, i believe we can train an auto-encoder on the training set and then passing eval or serving datasets and retrieving a reconstruction loss to determine if there is a skew.

Hi @jax79sg

you can get a look at this thread: