search

HsiehYiChia / Scene-text-recognition

Scene text detection and recognition based on Extremal Region(ER)
MIT License
163 stars 55 forks source link

The positive and negative samples cannot be uncompressed after downloading #9

Open ssifei opened 4 years ago

ssifei commented 4 years ago

Hi! Thank you for your project! Maybe the pos and neg is too large to unzip

I want to use text detection function without OCR,but I don't know how to prepare my own neg and pos. Could you tell me how to prepare training data ? Could you update the training data again ?

HsiehYiChia commented 4 years ago

Hi Most of the training data is collected from ICDAR data set, and categorize them in alphabetical order with case sensitive. You can download them from ICDAR website .However, latest ICDAR data set doesn't provide character training data, they now provide image with scene text, so you need you search old data set. I'm currently busy with work, and I'll update the training data again when available

SSif notifications@github.com 於 2019年11月16日 週六 下午11:35寫道：

Hi! Thank you for your project! Maybe the pos and neg is too large to unzip

I want to use text detection function without OCR,but I don't know how to prepare my own neg and pos. Could you tell me how to prepare training data ? Could you update the training data again ?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/HsiehYiChia/Scene-text-recognition/issues/9?email_source=notifications&email_token=ADGP7MMCLRMXGLTWNIANSKTQUAHNLA5CNFSM4JOFUR2KYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HZZOUZQ, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADGP7MIMGI3VRQJQ4ZFUXB3QUAHNLANCNFSM4JOFUR2A .

lyqlola commented 4 years ago

@HsiehYiChia You mentioned that your detection model is trained on character training data of ICDAR. Is is possible to train it on scene text dataset like Scene Text Detection on ICDAR 2015? image

HsiehYiChia commented 4 years ago

This would need lots of effort to make it work because our system is designed to train with character level. Therefore I don't think it is feasible

lyqlola notifications@github.com 於 2020年3月20日 週五 下午3:53 寫道：

@HsiehYiChia https://github.com/HsiehYiChia You mentioned that your detection model is trained on character training data of ICDAR. Is is possible to train it on scene text dataset like Scene Text Detection on ICDAR 2015? [image: image] https://user-images.githubusercontent.com/34086753/77145213-d819f180-6ac2-11ea-9681-c1abe44085d1.png

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/HsiehYiChia/Scene-text-recognition/issues/9#issuecomment-601573664, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADGP7MKLK6IAPTBJGK5QP43RIMOHVANCNFSM4JOFUR2A .