lartpang / PySODEvalToolkit

PySODEvalToolkit: A Python-based Evaluation Toolbox for Salient Object Detection and Camouflaged Object Detection
MIT License
168 stars 20 forks source link

关于曲线坐标的问题 #23

Closed LiuMingYy closed 2 months ago

LiuMingYy commented 1 year ago

非常感谢您杰出的工作!但是我对于F-measure和E-measure的横坐标有些问题,想请教下threshold是什么意思?为什么F-measure和E-measure会随着threshold的变化而变化?

lartpang commented 1 year ago

@LiuMingYy 因为这个工具对应的sod、cod类的任务,实际评估使用的预测是灰度图,而非二值图。而这些指标的计算使用的是二值图的形式。所以涉及到二值化的问题。也就涉及到了阈值选择的问题。

这样通过不同阈值(0~255共256个)划分的方式可以更直观的了解模型的真实预测的情况。 更好的预测应该具有更加二值化的预测效果,反映在曲线上,fmeasure、emeasure也应该更平直。

github-actions[bot] commented 2 months ago

This issue is stale because it has been open 7 days with no activity. Remove stale label or comment or this will be closed in 5 days.

github-actions[bot] commented 2 months ago

This issue was closed because it has been stalled for 5 days with no activity.