#visualrecognition

***Infogrex Offers Workable Free Webinar on Data Science***

Online Data Science Course showing some of the common disciplines that a data scientist...

.

Register Now www.infogrex.com

Contact : 9347412456, 040 - 6771 4400/4444

.

.

Today we presented at the Watson Developer Conference and shared the node.js code that you can use in your web app to automatically recognize video using Watson’s Visual Recognition service. 

We pass the video in the parameters to ffmpeg (video.mp4 in this example) and then we slice it into frames which then get sent to Watson’s Visual Recognition one by one. Once we’re done we just tally up the results!

NOTE: Make sure you have ffmpeg installed in your system

We’d like to thank the organizers of Watson Developer Conference for putting on such an amazing event and all the attendees for coming out and watching the presentation. 

Here is a screenshot of it in action:

Here is the Video that’s being processed:

By: Alex

For a recent Developer Day we put together a demo utilizing a custom made drone running off of a Raspberry Pi with the pi camera module. Using the drone we took a snap shot every several seconds and uploaded the images to IBM Cloud Object Storage, tagged the images using IBM Watson Visual Recognition, and stored the tags in a Cloudant database.

I left the drone camera pointed at me while I worked. It remotely uploaded a few dozen images with tags.

classes": [ { "class": "person", "score": 0.999994, "type_hierarchy": "/people" } ]

All of the images the drone takes are uploaded remotely and tagged in real time. We have a lot of opportunities with this method to then act on what the drone is seeing during the flight. And since the tagging is done using Watson and autonomous flight control software is improving, we have reduced the need for a human operator.

Here is a short example of how we are remotely storing and tagging the images. Note that I just assigned each image a random number for a file name. You can use any other method that works well for your use case.

function storePic(data, cloudant){ console.log('File loaded'); os.createContainer() .then(function(){ return os.setContainerPublicReadable(); }) .then(function(){ return os.uploadFileToContainer(Math.random()+'.jpg', 'image/jpeg', data, data.length); }) .then(function(file){ console.log('url to uploaded file:', file); tagPic(file, cloudant); return os.listContainerFiles(); }) .then(function(files){ console.log('list of files in container:', files); }); } function tagPic(url, cloudant){ console.log("tag called"); var params = {url: url}; visual_recognition.classify(params, function(err, res) { if (err) console.log(err); else{ console.log(JSON.stringify(res, null, 2)); var tagging_demo = cloudant.db.use('tagging_demo'); tagging_demo.insert(res, url, function(err, body, header){ if (err) { return console.log('[tagging_demo.insert] ', err.message); } console.log('You have inserted the doc.'); console.log(body); }) } }); }