The blobs (organelles/bacteria etc) sitting close to each other clump together and are counted as one. So, the number of object and their volume are not accurate. One way I solve that issue now is I reduce the supervoxel size when I do the final touching with manual mode. For example, I use bigger size (ie. 12, 14 etc) supervoxel and do the prediction as usual. I got the segmented blobs that touch each other. For the final touch, I change the supervoxel to a quite small size (i.e. 4,5 etc) and erase the small supervoxels between blobs that clump together. It is a bit tidious job, and there could be a better way to do it. And the other issue is, when I reduce the size of supervoxel to a smaller number, survos crashes sometimes (may be because the number of supervoxel becomes too high).
The blobs (organelles/bacteria etc) sitting close to each other clump together and are counted as one. So, the number of object and their volume are not accurate. One way I solve that issue now is I reduce the supervoxel size when I do the final touching with manual mode. For example, I use bigger size (ie. 12, 14 etc) supervoxel and do the prediction as usual. I got the segmented blobs that touch each other. For the final touch, I change the supervoxel to a quite small size (i.e. 4,5 etc) and erase the small supervoxels between blobs that clump together. It is a bit tidious job, and there could be a better way to do it. And the other issue is, when I reduce the size of supervoxel to a smaller number, survos crashes sometimes (may be because the number of supervoxel becomes too high).