Thank you so much for spending your time in answering my questions.
1) I read in the paper about the result on high resolution images and low resolution images. Can you help me figure out where high resolution and low resolution images are. Are there high resolution images in real and sim datasets and low resolution real and sim datasets?
2) During the computation of metrics, how do you ensure both the ground truth and the predicted images have the same number of masks?
There are high resolution images in the sim dataset, and both high and low resolution images in the real dataset.
There need not be the same number of predicted and GT masks; what happens during the metric calculation is that the predicted masks are matched to GT masks and then the metrics are calculated. I'd recommend checking out pycocotools for more info on how exactly this works.
Hi Sir,
Thank you so much for spending your time in answering my questions.
1) I read in the paper about the result on high resolution images and low resolution images. Can you help me figure out where high resolution and low resolution images are. Are there high resolution images in real and sim datasets and low resolution real and sim datasets?
2) During the computation of metrics, how do you ensure both the ground truth and the predicted images have the same number of masks?