Advances in computing capabilities combined with deep learning algorithms have given skilled machine learning teams, like NLP Logix, the ability to train computers to identify and classify images. Over the past two years, NLP Logix has invested heavily in the infrastructure to deliver world-class computer vision solutions.
This infrastructure includes the hardware and software required to develop, from end-to-end, a deep learning algorithm which can perform at or better than the ability of a human. The learning process includes the following steps and technology:
- gathering enough imagery samples to train an algorithm
- annotation software to highlight the imagery being learned
- deep learning algorithms combined with graphic processing units (GPU’s) to process the large amounts of data
- software to integrate with a customers system to process the imagery
- user interface to render model results
The rule of thumb at NLP Logix is “if a human can see it, then we can train a computer to see it just as well.” We’ve proven it time and again for clients as varied as healthcare providers to a large logistics company in North America.
NLP Logix trained an algorithm to identify signage along a rail network in North America
NLP Logix trained an algorithm to identify cancerous tumors in pathology slides
NLP Logix trained an algorithm to classify Her2 tumors