The previous two chapters have covered real use case implementation of NLP done through RNNs/LSTMs in Apache Spark. In this and the following chapter, we are going to do something similar for CNNs: we are going to explore how they can be used in image recognition and classification. This chapter in particular covers the following topics:
- A quick recap on what convolution is, from both the mathematical and DL perspectives
- The challenges and strategies for object recognition in real-world problems
- How convolution applies to image recognition and a walk-through of hands-on practical implementations of an image recognition use case through DL (CNNs) by adopting the same approach, but using the following two different open source frameworks and programming languages:
- Keras (with a TensorFlow backend) in Python
- DL4J (and ND4J) in Scala