There is a python and C++ api available for doing image recognition.The tutorial is provided in tensorflow.org but as tensorflow.js is released few months back only does it support all the api's available in the python and c++ implementation.
Vincent Mühler has created face-api.js, a JavaScript API base on tensorflow.js. You can find out the blog & code on the links below.
https://itnext.io/face-api-js-javascript-api-for-face-recognition-in-the-browser-with-tensorflow-js-bcc2a6c4cf07
https://github.com/justadudewhohacks/face-api.js
Adding to the answers above, tensorflow on JavaScript can be quite slow. Here's a quick comparison:
However, If you do run tensorflow.js on Node, you could make use of the binding directly to the TensorFlow APIs written in C which would be fast. You could also run the CUDA versions if you import the right packages on Node.js
On the browser WebGL is used to run tensorflow. Using the tensorflow or some ML on the browser, opens up whole new opportunities to do things from right within the browser.
As Jirapol suggested, you could take a look at https://github.com/justadudewhohacks/face-api.js which is super easy to use. It actually took a very short while for me to start writing a facial recognition login system on node using face-api.js. Here's a link if you want to take a look at the unfinshed code: https://github.com/WinstonMarvel/face-recognition-authentication
does it support all the api's available in the python and c++ implementation.
No, it still has a limited amount of features. Keep in mind it still in version 0.11.6 so that will change. You can look at the documentation to see what's available.
If you want to port a specific model to tfjs try to get it as a keras model then convert it using tensorflowjs_converter to a tfjs compatible one like this tutorial shows.
There is even a tfjs examples which works with webcam data (Tutorial, Live Demo), so you could look into that to start.
Yes it can.
and also with the help of webassembly and SIMD in the browser.
you can have a smooth experience of image processing and video processing in the browser.
have a look at this link from google v8.
the good news is that with the same api you can run Tensorflow.js in the browser, node.js and React Native all with the native speed and using native capabilities
Related
I have a question on using tensorflow on google cloud platform.
I heard that Google cloud tensorflow doesnt support Keras (keras.io). However, now i can see that Tensorflow has its own API to access Keras (https://www.tensorflow.org/api_docs/python/tf/contrib/keras).
Given this, can I use the above mentioned API inside google cloud, since it is coming out along with Tensorflow package? Any idea sir?
I am able to access this API from the tensorflow installed on a anaconda machine.
Option 1# Please try package-path option.
As per the docs...
-package-path=PACKAGE_PATH
"Path to a Python package to build. This should point to a directory containing the Python source for the job"
Try and give a relative path to keras from your main script.
More details here:
https://cloud.google.com/sdk/gcloud/reference/beta/ml-engine/local/train
Option 2# If you have a setup.py file
Inside your setup.py file within setup call pass argument install_requires=['keras']
Google Cloud Machine Learning Engine does support Keras (keras.io), but you have to list it as a dependency when starting a training job. For some specific instructions, see this SO post, or a longer exposition on this blog page. If you'd like to serve your model on Google Cloud Machine Learning or using TensorFlow Serving, then see this SO post about exporting your model.
That said, you can also use tf.contrib.keras, as long as you use the --runtime-version=1.2 flag. Just keep in mind that packages in contrib are experimental and may introduce breaking API changes between versions.
Have a look at this example on git which I saw was recenly added:
Keras Cloud ML Example
I'm working on an app written in Codename One together with the parse4cn1 library, the combination of which is a real pleasure to use. However, I need support for a few things in parse4cn1 that are not implemented, most importantly ACL and was wondering if Chidiebere has any hints on how to do this (e.g. how did you implement parse4cn1 yourself - from scratch or copying the open source Parse SDK for Android)? If I manage to do something of a decent quality I will try to share back. Thanks in advance
I never got around implementing ACLs (it's still on the TODO list). parse4cn1's interface closes resembles the Parse Android SDK interface and I'll like it to stay that way for convenience. In this case, the interface of interest would be the ParseACL which is documented here.
The actual implementation will need to be done via REST API calls.
Things to bear in mind:
We use the Android SDK API simply for defining methods and signatures for the corresponding class in ParseACL but do not use the SDKs for anything can be be done via REST.
By design, any calls requiring the master key will not be supported in parse4cn1 due to security considerations. If really needed, the functionality should be exposed via server-side cloud code.
Pull requests without unit tests for the added functionality or breaking existing tests will be rejected.
See also the Contributions section of the parse4cn1 github repo.
Good luck with your implementation and I hope to see a PR from you soon ;)
It was implemented from a Java port on top of the REST API's here but was later modified to use the SDK's to allow things like push (which are now no longer relevant).
In the past I just contributed pull a request to the project to get the fixes/features I needed. It was really easy to work with and compile.
I've written an AIML file for a chat bot and I'd like to build an interactive web application which allows me to chat with the bot in the web browser.
Is it possible to achieve this with HTML & Javascript?
There is no short answer on how to write a web application which allows a user to interact with your AIML. Writing such an application from scratch will be much more work then compiling the AIML was.
The easiest option would be to use a pre-built service like PandoraBots which allows you to upload AIML files and interact with them in the web browser. It's free to use the explorer part of website. They also have paid developer options which generates an API to bridge your AIML script and any applications you might want to build. It can be easily connected to work with common chat apps like Google talk ect.
If you decide to build everything from scratch you might want to check out the AIML Interpreter library for nodejs.
UPDATE: Here is a node.js based interpreter that you might find useful https://github.com/mrchimp/surly2
I was looking at AIML too and had similar questions. I just found RiveScript RiveScript and it looks like it fits your need to run javascript based on a match. It is not AIML, but very close. There is also at least one tool to convert from AIML to RiveScript, so I would say this fits your needs within those constraints.
We have a library with very complex logic implemented in C. It has a command line interface with not too complex string-based arguments. In order to access this, we would like to wrap the library so that it can be accessed with simple XML RPC or even straightforward HTTP POST calls.
Having some experience with Java, my first idea would be
Wrap the library in JNI/JNA
Use a thin WS stack and a servlet engine
Proxy requests through Apache to the servlet engine
I believe there should already be something simple that could be used, so I am posting this question here. A solution has the following requirements
It should be deployable to a current linux distribution, preferrably already available via package management
It should integrate with a standard web server (as in my example Apache)
Small changes to the library's interface should be manageable
End-to-end (HTTP-WS-library-WS-HTTP) the solution should not incur too much overhead, but reliability is very important
Alternatively to the JNI/JNA proposal, I think in the C# world it should not be too difficult to write a web service and call this unmanaged code module, but I hope someone can give me some pointers that are feasible in regards to the requirements.
If you're going with web services, perhaps Soaplab would be useful. It's basically a tool to wrap existing command line applications in SOAP web services. The web services it generates look a bit weird but it is quite a popular way to make something like this work.
Creating an apache module is quite easy and since your familiar with xmlrpc you should check out mod-xmlrpc2. You can easily add your C code to this apache module and have a running xmlrpc server in minutes
I think you may also publish it as a SOAP based web service. gSoap can be used to provide the service interface out of the library. Have you explored gSOAP? See http://www.cs.fsu.edu/~engelen/soap.html
Regards,
Kangkan
Depends what technology you're comfortable with, what you already have installed and working on your servers, and what your load requirements are.
How about raw CGI? Assuming the C code is stateless between requests, you can do this without modifying the library at all. Write a simple script which pulls the request parameters out of the CGI environment, perhaps sanitises the input, calls the library via the command-line interface, and packages the result into whatever HTTP response you want. Then configure Apache to dispatch the relevant URL(s) to this script. Python, for example, has library support for XML-RPC, and so does every other scripting language used on the web.
Servlets sound like overkill, but for instance if you want multiple requests per CGI process instantiation, and don't feel like getting involved in Apache configuration, then it might be easiest to stick with what you know.
I'm doing a similar thing with C++ at the moment. In my case, I'm writing a PHP module to allow PHP scripts to access the logic in my C++ library.
I can then use whatever format I want to allow the rest of the world to see it - initially it will just be through a PHP web application but I'll also be developing an XML-RPC interface.
If you're going down the JNI route, check out SWIG.
http://www.swig.org/Doc1.3/Java.html
Assuming you have headers to project bindings with, swig is pretty easy to work with.
I am interested in knowing if there are any server-side web application frameworks which integrate nicely with CouchDB? Does anyone have any experience in doing this? It seems like a dynamic language would be well-suited for playing with the JSON, but I am more interested in hearing about how it would fit in with the framework and the application's design.
Two frameworks that I would suggest for CouchDB are Ruby on Rails and Django. Both have a small file you can include that allows for easy interaction with CouchDB. For Ruby/Rails, this gives you the ability to write code that looks like this (code snippets yanked from here):
# Create the database
server = Couch::Server.new("localhost", "5984")
server.put("/foo/", "")
# Insert a new document into the database
doc = <<-JSON
{"type":"comment","body":"First Post!"}
JSON
server.put("/foo/document_id", doc)
# Get the document back later
res = server.get("/foo/document_id")
json = res.body
puts json
Python/Django lets you do the same with a relatively minimal amount of work (see here). Both of these aren't at the web framework level but they require a minimal amount of work to set up and are pretty easy to get going in Rails and Django. The Django approach still requires some packages to be installed so if you just can't do that for some reason the Rails approach is the way to go.
Another good how-to on Python on Django can be found here (also lifted from the CouchDB FAQ).
The only web framework that dedicates itself to CouchDB is currently CouchDBKit for Python.
Check out the official wiki page that lists how to get started in your language:
http://wiki.apache.org/couchdb/Basics
Pick the language and framework that suits you best and then use one of the light CouchDB libraries with it.
It seems that things are move quite quickly at the moment for CouchDB. I'm sure there will be more frameworks out there soon with CouchDB support. I'm currently looking into building one for PHP.
I have had good success with jcouchdb for Java and CouchApp for JavaScript and CouchDBKit with Python. All of these are actively developed, open source and well designed and easy to enhance if they are missing something you really need. I have submitted patches and feature enhancements for jcouchdb and couchapp both.
Actually, you don't really need such a framework. Instead, you can just write the whole web application in CouchDB. It allows you to generate HTML files, or any other XML derived format, and you can even use HTML-templates. I consider this a good choice, because JavaScript is a rich and flexible language. On the other hand you don't have the overkill of a connection between the database and your web application.
For more details, check out: http://books.couchdb.org/relax/design-documents/shows
There's also a related question: Using CouchDB to serve HTML
Depending on what you want to build CouchApp may be something to look at: It's specially designed for writing apps with CouchDB:
https://github.com/jchris/couchapp/wiki/manual