Comments (6)
Nice question. I am wondering the same. Please tell me how to train the model?
from handwriting-ocr.
Hi,
This is question depends on the model you want to train. All the notebooks that are for model training contain name Classifier. These notebooks load data from the data folder (if you didn't already, you have to download the data from provided URL), process them and train the model which is then saved in the models folder.
You don't have to do much more than replace the original date with yours and train the model. Your data have to be in the right format which depends on the type of a model.
Often the data are stored as and image file with name in the format: label_timestamp.jpg
.
If you need more details, please specify the model you want to train.
from handwriting-ocr.
I wanted to train the word-classifier CTC.
How to do it?
from handwriting-ocr.
OK, that's the easy one.
The training code is in this notebook: WordClassifier-CTC.ipynb. Currently, the data are loaded from folder data/words2/
(the location is parametr of loadWordsData()
). In this folder I have images of words which are already normalized (grayscaled and with height: 60px). The words' images are named as word_timestamp.jpg
(word stands for correct label and timestamp can be just random number).
For example, following image is named as sell_15132719.jpg
:
The loadWordsData()
loads grayscaled images and outputs numpy array of images and labels. The model is then trained and output into location defined by save_location
variable.
I hope this helps.
from handwriting-ocr.
What are the .txt files in data/words2? I am going to retrain the char classifier and it needs the .txt files. How can I generate .txt files for my data?
from handwriting-ocr.
This question is duplicate with #44
from handwriting-ocr.
Related Issues (20)
- Query: Punctuation Marks HOT 1
- Language HOT 3
- not giving output same as in your github ocr.ipynb ctc model HOT 9
- ValueError: zero-size array to reduction operation minimum which has no identity
- unimplementederror: tensor array has size zero, but element shape [?,256] is not fully defined. currently only static shapes are supported when packing zero-size tensorarray
- File models/gap-clas/CNN-CG.meta does not exist.
- No Function : imageNorm ? HOT 1
- 'TrainingPlot' object has no attribute 'updateCost' HOT 2
- Tensor shape error / not training my images HOT 1
- handwriting-ocr/word_classifier_CTC.ipynb question
- ModuleNotFoundError: No module named 'ocr'
- ValueError: too many values to unpack (expected 2) HOT 5
- training time
- How much time it takes for training i am waiting for 2 hours and what is value of LOSS_ITER and also can you check the train.csv, dev.csv, test.csv i have generated are good to use or have some error?
- What does this code doing and how can i visualize it's output. HOT 1
- ValueError: Cannot feed value of shape (13, 1, 3600) for Tensor 'inputs:0', which has shape '(None, 64, None, 1)'
- Javascript implementation HOT 1
- File does not exist. Received: F:\MY_PROJECT\handwriting-ocr-master\src\ocr\../../models/gap-clas/CNN-CG.meta. HOT 1
- Request for resources
- field to access
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from handwriting-ocr.