search

GlassyWing / text-detection-ocr

Chinese text detection and recognition based on CTPN + DENSENET using Keras and Tensor Flow，使用keras和tensorflow基于CTPN+Densenet实现的中文文本检测和识别
Apache License 2.0
285 stars 116 forks source link

What is the training Dataset you used for DenseNet +CTC text recognition? #7

Open WrathofBhuvan11 opened 5 years ago

WrathofBhuvan11 commented 5 years ago

...

GlassyWing commented 5 years ago

The dataset generated from a lot Chinese corpus, It contains three parts:

  1. images (32 x 280)
  2. train.txt, test.txt (Each line split by image path and sentences)
  3. dictionary file (character to index) you can fetch it from https://github.com/YCG09/chinese_ocr
WrathofBhuvan11 commented 5 years ago

Well I have trained CRNN on synth90K dataset... For English.. by it won't recognise multiple words in a single segmented ROI

On Mon 21 Jan, 2019, 8:47 AM pure glay <notifications@github.com wrote:

The dataset generated from a lot Chinese corpus, It contains three parts:

  1. images (32 x 280)
  2. train.txt, test.txt (Each line split by image path and sentences)
  3. dictionary file (character to index) you can fetch it from https://github.com/YCG09/chinese_ocr

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/GlassyWing/text-detection-ocr/issues/7#issuecomment-455936251, or mute the thread https://github.com/notifications/unsubscribe-auth/AijbZfmvZ_g4TUzVtHuazMaYgtnfzrEeks5vFTFOgaJpZM4aJs_o .