Is Pre-training Applicable to the Decoder for Dense Prediction?

Open in new window