What can human minimal videos tell us about dynamic recognition models?

Re-assign community