Hello All 😊. After working with the Face to Voice for visually challenged and tried to make it as a product. But a question arises inside me.✋. Recognizing a human is fine. But what if other objects come around them?😅. Simple, they ask the person near around them? Yuk! That's not going to happen when the world have a electronics hobbyist like me. So that I've created a system which can differentiate the object in real time and most importantly it should be portable.
As always Raspberry Pi is a heart of the system with the Raspbian Jessie is used. Thank to Google tensor-flow API which is a open source library for Machine Learning, they've COCO - Common Object in Context. Lots of models set are available in the tensor-flow. I just utilized a basic model sdd_mobileclient_v1, for testing it out the performance. And I just wondered with it's speed😭. Further, I'm working on to make it as a perfect product.
Working Images:
Pretty Good!
I'll work on to improve the speed of processing on raspberry pi and embedding a audio out to it. Bye for now!✌
Have you looked at the Movidius Neural Compute Stick. It'll interface through USB, and will accept tensorflow programming.
ReplyDeleteNice Article Bro
ReplyDeleteVisit Tech News to get tips and tricks of Technology, Android, Websites and others.