search

ZJULearning / pixel_link

Implementation of our paper 'PixelLink: Detecting Scene Text via Instance Segmentation' in AAAI2018
MIT License
768 stars 255 forks source link

Can you provide the setting that reaches to 85.x% accuracy? #48

Open GodOfSmallThings opened 6 years ago

GodOfSmallThings commented 6 years ago

@dengdan I tried to reproduce the result that you've reached, which is 85%. However, I failed to make that result. It goes as much as 83% in my experiments. Can you provide the setting(e.g hyper parameters that you've used for pretraining and finetuning) using both Synth Text and ICDAR datasets?

Thank you.

kulkarnivishal commented 6 years ago

Hi could you provide the setting for 83%. I am using COCO dataset and I am unable to reach even 80. Have been training for 3 weeks

mn7142 commented 6 years ago

@GodOfSmallThings @kulkarnivishal please help me to evaluate my result how can you calculate your result?

CXY573 commented 5 years ago

how can you calculate your result?

XiaoYigwr commented 5 years ago

@GodOfSmallThings @kulkarnivishal @mn7142 How did you test the results of the model?你们是阿如何检验结果的 Could you share the script?可以分享一下脚本吗 I don't konw how to do!

GodOfSmallThings commented 5 years ago

@XiaoYigwr Go to http://rrc.cvc.uab.es/?ch=4&com=mymethods&task=1 after login. There is standalone evaluation script and it is easy to use.

PeterStone007 commented 5 years ago

standalone evaluation script is not available. do you have other source?