Recently saw the talk in meta events about self-supervised learning by @pdollar. I hope this is the project that was mentioned in the talk.
I've been using detectron2 for instances segmentation for some time now - they give incredible results.
I want to improve 1. Accuracy and 2. Fairness of our predictions by building a foundational model on hundreds of thousands (if PoC is a success, will use millions) of custom images of India and integrate with detectron2.
Would you be kind enough to guide me throughout the journey.
Hi,
Recently saw the talk in meta events about self-supervised learning by @pdollar. I hope this is the project that was mentioned in the talk.
I've been using detectron2 for instances segmentation for some time now - they give incredible results.
I want to improve 1. Accuracy and 2. Fairness of our predictions by building a foundational model on hundreds of thousands (if PoC is a success, will use millions) of custom images of India and integrate with detectron2.
Would you be kind enough to guide me throughout the journey.
Regards, Rakesh