Why the hmean is lower than EAST

[xxlxx1]

https://github.com/argman/EAST train in EAST the hmean is 0.80 the precision is 0.846 the recall is 0.77, the detection part of FOTS is vary similar to EAST, and the result of full FOTS in icdar website result is better than EAST. I think the best perfomence of FOST should be better than EAST, maybe you can improve your FOTS model.

[Vipermdl]

Thanks for your advice, the code isn't completed. For example, a random seed has a big impact on the results.