Example: bankruptcy

Masked Feature Prediction for Self-Supervised Visual Pre ...

frame patches into tokens, as explored in BEiT [2,73]. However, this requires an external tokenizer which can be limited in compute-intensive video understanding scenario. We present Masked Feature Prediction (MaskFeat), a pre-training objective that directly regresses features of the masked content. Specifically, our approach ingests the

Tags:

  Bites

Information

Domain:

Source:

Link to this page:

Please notify us if you found a problem with this document:

Other abuse

Advertisement

Transcription of Masked Feature Prediction for Self-Supervised Visual Pre ...

Related search queries