I am very new to the course, and am just starting.
Came across this announcement from Meta.
Instead of asking my friend GPT I thought to ask this forum of esteemed practitioners. Is there anything in there that could simplify or anyhow affect our learning of computer vision?
Thank you
I guess the impact of SAM will be similar to other foundational models. For segmentation tasks in images it has been trained on (so mostly natural photographs) it will probably be great. For everything that is not cookie cutter standard it will be worse than a custom trained segmentation model. Examples would be medical images like pathological slides or MRIs and CTs.
In my view, data annotation appears to be the most impactful approach. It means that the performance of AI is intricately linked to the quality of the labeling process and the factors considered. This task presents several challenges, including coco annotator, yet with a skillful approach, one can achieve impressive outcomes.