I've already read everything around ml kit tutorial https://firebase.google.com/docs/ml-kit/android/recognize-text . I'm interested only in local ocr not cloud version. Of course as question suggests it's about text recognition.
There is no option for you to control the text color that the ML Kit model looks for. You will either have to perform filtering in your app after calling ML Kit, or train your own model for that.
Related
We want to embed a 3D CAD viewer into our web-based collaboration platform. It would work similar to the one offered by Opencascade. You can see it here.
We do not charge for the use of our platform and we would prefer an open source option if it is available (Opencascade viewer is not open source).
Does Autodesk Forge have a similar offering?
Thanks!
CAD viewer solution for a WEB can be split into several pieces:
On-server converter of CAD model.
Conversion of CAD file into intermediate format. You might need a lot of Data Connectors in case if you need supporting wide range of CAD formats, or restrict functionality to import vendor-neutral formats like STEP.
Generation of Web-friendly triangulation-based 3D model. This step might include model simplification in case if you have to deal with large models.
Server backend implementing business logic and backend for 3D viewer. The complexity of 3D Viewer backend depends on functionality provided with it and ability of viewer to handle extra-large models. In simplest case, this might be just a file server providing 3D model in format like glTF.
Client part, e.g. HTML+JavaScript code for displaying 3D model using WebGL, as well as GUI and application-specific logic. The viewer might just display 3D model (in which case it is barely distinguishable from non-CAD viewers), or give many features like Shaded and Wireframe (not just mesh edges) modes, picking sub-parts, Clipping Boxes, clipping with capping, annotations, dimensions, explode view, HLR view, a bunch of them!
Although it is technically possible loading some small CAD models directly in the Browser, in general this task is expected to reside on server to improve user experience and give some protection to intellectual properties (of CAD models). The converter may be based on Open CASCADE Technology, which is an open source framework, in case if STEP/IGES input file formats are enough (output could be your custom file format or glTF 2.0, for instance). But in case of native CAD formats support (like CATIA and others) you will unlikely find any free open-source libraries.
The Web viewer itself relies on WebGL and normally can be based on any traditional JavaScript-based library like Three.js (or existing 3D Viewer on top), or may rely on C++-based engine (again, Open CASCADE Technology can be build as WebAssembly module). WebAssembly-based viewer may give some performance benefits and it is more friendly to C++ developers (but apparently - less friendly to JavaScript developers). There is a hidden issue, though, as many open-source C++ libraries commonly have LGPL license - pretty good for desktop, but might have legality issues in case of a Web application. In contrast, WEB-based libraries are normally more proprietary-friendly with permissible licenses ignoring open-source software ideas put into licenses like (L)GPL.
In general, 3D CAD Viewer definition is very wide, and requirements list in your question are very basic and unclear - just to embed 3D Viewer in your WEB application. You may write your own Web Viewer based on open source components, or consider commercial options already implementing features you are looking for, and aware of / providing some solutions for handling tricky problems.
Looks like Forge Viewer fits the bill perfectly and covers most of the features available with the demo ;)
You may experience/preview Viewer following this link using a model of your own here and follow here to build up a MVP here...
For the features that are not readily/built-in out of the box Viewer is highly extensible and offers API making it pretty straight forward to develop extensions following here and also the blogs/tutorials available here.
I have learned the overview and functionality of Siamese network as it is One Shot Learning, I have tried some link from web which can give the demo of face detection using Siamese.
But I am not able to find any demo in which all the steps like training dataset, verification and testing steps are there. Can anyone have any reference link in which I can just detect the face from camera and comapre it with my Training dataset and that functionality I want using Siamese whether it is using tensorflow or keras.
Here is an example what you are looking for. The author has explained very nicely with python script, to find the similarity between ships and ice-berg, which can be easily expandable to any other application.
I have deployed a regression model on azure ML , is it possible to get the model weights/coefficients of the model programatically from azure, rather than getting predicted value? .
I think you can do so, in your training experiment add an output to your evaluate model module then select deploy webservice right away without going through the predictive experiment option.
Once You publish and click the TEST button You should the values as below
No. Currently we do not feature exporting weights from the models including with Azure Machine Learning.
If you have a method for extracting weights from Python models, you may be able to work this out using the execute Python Script module.
The primary purpose of Azure Machine Learning is to make deployable and scalable web services from the machine learning modules. Though the authoring experience for creating ML models is great, it is not intended to be a place to create and export models, but instead a place to create and operationalize your models.
UPDATE New features may make this answer outdated.
I am creating a software for calculating gearbox parts using Swing and OpenGL (with JOGL).
With this software the calculated parts are shown in 3D. The following picture shows an example part:
At first the geometry is specified/calculated (tip/root Diameter, width, number of teeth...) and out of this data the 3D-part is created. This example part is made with JOGL-Code because it's easy to use for me and can be integrated in a Swing-GUI.
Now a played around a bit with JavaFX Application GUI's and decided to continue making my software in JavaFX because it offers everything i need, looks much better and so on. The problem is I can't use my JOGL-OpenGL graphics.
So here are the possibilities for me:
Search for a way to integrate OpenGL-graphics in a JavaFX-Application - but I googled a bit and this seems to be quite difficult and the possible solutions seem to be very slow.
Use JavaFX integrated 3D-graphics (TriangleMesh) - but this is difficult to understand and I can't find good examples to learn it.
Any other solution?
Can you please help me finding a good solution? What would you do? My favorite would be to continue using OpenGL, but is there any good solution for this yet?
If you are looking for some JavaFX 3D resources that may help you decide whether its already mature or not, these are available, among others:
3DViewer here from the OpenFJX project.
InteractiveMesh importers and browser.
F(X)yz, a new JavaFX 3D library that provides additional primitives, composite objects, controls and data visualizations that the base JavaFX 8 3D packages do not have.
Multimodel3DFX, repository contains the code for the example shown on the article "Building Castles in the Sky. Using JavaFX 3D to model historical treasures" published in Oracle Java Magazine, issue from November/Dicember 2014.
Using a PolyLine3D from F(X)yz and the Multimodel3DFX plaftorm I was able to extrude a 2D polyline of a spur gear, so you can see how it's rendered in a JavaFX 3D subscene:
These are also mathematical 3D models generated using the F(X)yz library, all of them based on TriangleMesh.
As you can see, JavaFX 3D API, with its pros and cons, can be easily extended to achieve complex tasks.
I have succesfully trained tesseract 2.x to recognize a few specific fonts. However, it seems that I can't make tesseract to recognize all of those fonts at the same time - i.e. source image contains all of them. Currently, only one set of tesseract data can be put into tessdata folder (i.e. one set with one trained font).
I know that tesseract 3.x handles correctly multiple fonts - however, I can't upgrade, since there's no decent binding to .NET, that has same features as .NET binding of version 2.x.
Also, I would like to avoid doing all the preprocessing and OCR itself several times, for each font.
For Tesseract 2.0x, a language data pack can recognize multiple fonts. Did you cluster your training files?
There are a couple excellent .NET wrapper for Tesseract 3.01. Check its AddOn page for more info.