search

tonghe90 / textspotter

324 stars 112 forks source link

Will you release the training code? #8

Closed smilewsw closed 6 years ago

tonghe90 commented 6 years ago

@smilewsw , Thanks for your interest in our work. For some IP issue, I don't intend to release the training code. If you have any question, please feel free to contact me.

wanghaisheng commented 6 years ago

@tonghe90 can you improve your existing trained model using ctwdataset and ICDAR2017 Robust Reading Challenge on COCO-Text and [Overview - ICDAR2017 Competition on Multi-lingual scene text detection and script identification]()

Jacoobr commented 6 years ago

@tonghe90 Hello tonghe90, Thanks for your amazing work. I'm a fresh researcher on text detection and recognition work. Now, my work is to detect and recognize the text on the logo of many shops.Your work will inspire me a lot to my task. But I'm confused with the step of training the model with the method you mentioned with my own data set. The problems are as follows: 1) How can I use (or organize) the network structure of test_iou.pt、test_lstm.pt that you provided when I training my own model which i plane fine tune with your textspotter.caffemodel that you trained. What's more, would you mind provide me the training network structure (train_val.pt) that you used in your paper. There is my email address: xiaojianli@stu.xjtu.edu.cn. Your work will help me a lot to my research. Look forward to your favourable reply. Thanks.

tonghe90 commented 6 years ago

Hi @Jacoobr I will release the training.pt someday, but I am now busy other tasks. So it may be this weekend