DOVER指标怎么比1还大？

[Zannnn]

你们不是使用的https://github.com/VQAssessment/DOVER中的源码直接计算的么？源码中
def fuse_results(results: list):
x = (results[0] - 0.1107) / 0.07355 * 0.6104 + (
results[1] + 0.08285
) / 0.03774 * 0.3896
print(x)
return 1 / (1 + np.exp(-x))
保证了结果在0-1之间啊

[IceClear]

In our evaluation, we simply use the technical score here.
We only focus on the quality regarding degradations, so we do not consider aesthetic scores.
We use the score numbers directly shown in the file, which is amplified by 100.