Open qxr04025 opened 4 years ago
The mask was build with respect to ground truth results, using hole filling on expert segmentations.
@afabijanska Thanks for your reply! But when we want to segment a cell image by the proposed method, we do not have its ground truth. In that case, how can I find the border area and set the border to zero, and then get final segmented result and calculate the cell morphometric parameters?
Probably you could threshold & skeletonize the segmentation result, set image border pixels into 0 (let say 2-3 pixel frame) and spur the result.
Hello, I want to know how to get the 2_mask.png, which you use to postprocess the U-net prediction 2.png to get the final result.