FastSAM
FastSAM copied to clipboard
Training Cost
Hi! Thanks for the great repo! I really like it. Could you please provide more details about training costs, such as GPUs / training time? Besides, why are only 2% of images from SA-1B used for training? Is there some reason for the setting?
Hi, thanks very much for your attention. 2% of the images are randomly selected for basic validation purposes. We plan to release the training and finetuning codes in the near future. Also, we believe more training data leads to better performance. As for the training details, YOLOv8 is the reference since the first stage is based on it.