Description
#6 PyData Warsaw
Images form a significant part of data. Yet, until recently, it was difficult to programmatically extract metadata about images. Now, thanks to TensorFlow, we are able to develop algorithms that can give us information about images. In this presentation, Yaz will show us how to quickly get set up with TensorFlow using containers. To be even more efficient, what is becoming known as transfer learning will be demonstrated. An existing image recognition model will be used rather than the time-consuming approach of building one from scratch. Subsequently, this classifier model will be trained with new images. And finally, there will be a live test of this model applied to random new images.