Comments (4)
Shreeshrii
commented on June 8, 2024
That was an experimental traineddata. I haven't really used it.
You can try fine-tuning eng if you have some specific emojis that you want
to recognise.
…On Wed, Jun 17, 2020, 09:33 Avani Sharma ***@***.***> wrote:
Hi @Shreeshrii <https://github.com/Shreeshrii>, I am using
emoji.traineddata from emojilegacy folder to recognize emojis and text in
Whatsapp/Facebook Chat Images. So while recognizing I am using this command
:
pytesseract.image_to_string (Image.open('guguemoji.jpeg'),
lang=("emoji+eng")). However the performance is very poor. It could not get
a single emoji right. But detects english language text very well. I think
I am missing something on combining two models, or will I need to fine tune
/ cut off the few top layers from english model in tessdata_best myself to
retrain it for emoji data? Can you please suggest a possible way of atleast
increasing the accuracy for ("emoji + eng").
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABG37IZUIMRW2ZBQKSL4ZQLRXA6BJANCNFSM4OAGSHJQ>
.
from tessdata_emoji.
Avani1994
commented on June 8, 2024
Hey Thanks, @Shreeshrii, I just wanted some guidance from you, so we can create custom tif and box files using tesseract 3 and can train a few specific emoji, but with LSTMs can only use synthetic data generated using text2image program. I have following questions:
-
What do you think should we be using for our use case, should we go with basic model or should go with lstm one. What do you think would be more accurate?
-
Also if we want to go with Tessercat 4 (Lstms), can you please guide on from how should I create emoji training data, like should we take pics for emoji (whatsapp chat emoji images (one per text line)) and then create corresponding box files using lstmbox as we cant use tesstrain.sh because we don't have any language as emoji (--lang emoji) in base models.
-
What I am more concerned is since Deep Neural nets require a lots of data to get trained and produce some good results, Will it be possible that we fine tune it with few specific emojis and it will recognize them accurately? Basically I want to know how much train data will be needed to fine tune english model to be able to recognize some specific emojis. If you could suggest some starting point it will be helpful.
from tessdata_emoji.
Shreeshrii
commented on June 8, 2024
Language code doesn't matter, you can do training as eng and at end create
traineddata with a different name.
One character per line won't give very good results with lstmtraining.
It is possible that tesseract 3 or legacy engine will give better results
for this. You have to try it out. Legacy training doesn't take too long.
Have you seen tesstrain repo. Using a makefile you can train using single
line images and their groundtruth.
…On Wed, Jun 17, 2020, 11:58 Avani Sharma ***@***.***> wrote:
Hey Thanks, @Shreeshrii <https://github.com/Shreeshrii>, I just wanted
some guidance from you, so we can create custom tif and box files using
tesseract 3 and can train a few specific emoji, but with LSTMs can only use
synthetic data generated using text2image program. I have following
questions:
1.
What do you think should we be using for our use case, should we go
with basic model or should go with lstm one. What do you think would be
more accurate?
2.
Also if we want to go with Tessercat 4 (Lstms), can you please guide
on from how should I create emoji training data, like should we take pics
for emoji (whatsapp chat emoji images (one per text line)) and then create
corresponding box files using lstmbox as we cant use tesstrain.sh because
we don't have any language as emoji (--lang emoji) in base models.
3.
What I am more concerned is since Deep Neural nets require a lots of
data to get trained and produce some good results, Will it be possible that
we fine tune it with few specific emojis and it will recognize them
accurately? Basically I want to know how much train data will be needed to
fine tune english model to be able to recognize some specific emojis. If
you could suggest some starting point it will be helpful.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABG37I4GJLS2NEAVTFQ4XRTRXBPBVANCNFSM4OAGSHJQ>
.
from tessdata_emoji.
Avani1994
commented on June 8, 2024
Thanks @Shreeshrii will definitely try with Tesseract 3. However, for Tesseract 4 I saw tesstrain repo. Though seems like providing groundtruth for line images is a big task for emojis as tesseract currently does not detect emojis well. So other than doing it manually, are you aware of any other approach? Since you have already done emoji training, it will be helpful to know how you came up with ground truth for emoji line images.
from tessdata_emoji.
Related Issues (2)
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 tessdata_emoji.