Machine Learning

HOW IT WORKS

When Machines Learn

Machine Learning projects can utilize a wide range of learning algorithms from a simple SVM to deep neural networks. A simple ML project involve a few standard steps: data collection, model selection, training and deployment.  

Gather

Data centric machine learning is becoming increasingly popular. Data quality is shown to be far more important than any other factors in a successful ML product.  

Train

Training is a continues process. Model maturity depends on continues training. 

Deploy

A successful product not only satisfies training metrics such as accuracy, but it also needs to pass availability, reliability, security and performance test. A great deployment creates a great experience.  

PORTFOLIO

How it works

Below are a few Machine Learning projects presenting the ML production process explained above from start to finish. All the ML apps are cloud-based and accessible via Restful APIs. 

Instance Segmentaion

Model trained to segment USB cables and wires in an image.

Picture Swap

Replace a picture with another using image segmentation model.

Object Detection

Model trained to detect sofa and picture frames. 

check out the demo here!