The model is open source and is pre-trained on 142 million images in self-supervised fashion without any labels

The post Meta Unveils DinoV2, A Self-Supervised Vision Transformer Model appeared first on Analytics India Magazine.