williamlzw / microocr.net Goto Github PK
View Code? Open in Web Editor NEWText recognition model using TorchSharp.
License: MIT License
Text recognition model using TorchSharp.
License: MIT License
I found that in this project, the accuracy of identifying 1 to 4 characters at one time is relatively high. If the number of characters increases, no matter how you train, the output results will have more problems.
public List<Tuple<string, List<float>>> Decode(torch.Tensor preds, bool raw = false)
{
preds = torch.nn.functional.softmax(preds, 2);
var ret = preds.max(2);
var preds_idx = ret.indexes.detach().to_type(ScalarType.Int32).to(CPU);
var preds_score = ret.values.detach().to(CPU);
List<Tuple<string, List<float>>> retList = new List<Tuple<string, List<float>>>();
for (int i = 0; i < preds_idx.shape[0]; i++)
{
var word = preds_idx[i].data<int>().ToArray();
if (raw)
{
var scoreList = preds_score[i].data<float>().ToArray();
var score = scoreList[0];
string retStr = "";
List<float> retScoreList = new List<float>();
for (int j = 0; j < word.Length; j++)
{
var charIdx = word[j];
retStr = retStr + _idx2char[charIdx];
}
retScoreList.Add(score);
var tuple = new Tuple<string, List<float>>(retStr, retScoreList);
retList.Add(tuple);
}
else
{
var scoreList = preds_score[i].data<float>().ToArray();
string retStr = "";
List<float> retScoreList = new List<float>();
for (int j = 0; j < word.Length; j++)
{
var charIdx = word[j];
if (charIdx != 0 && !(j > 0 && (word[j - 1]) == charIdx))
{
var bb = charIdx;
retStr = retStr + _idx2char[bb];
var score = scoreList[j];
retScoreList.Add(score);
}
}
var tuple = new Tuple<string, List<float>>(retStr, retScoreList);
retList.Add(tuple);
}
}
return retList;
}
Your project is excellent
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.