Convolutional Neural Networks – Semantic Segmentation – Hands-on Session

Convolutional Neural Networks – Series of 4 Sessions
1) Classification – completed
2) Semantic Segmentation – Apr 30th
3) Object detection – May
4) Object localization – June

Agenda – Semantic Segmentation
1. Recap of the previous meetup – CNN workings
2. Semantic Segmentation and its applications in various industry.
3. Approaches for developing models for semantic segmentation.
4. Hands-on session – Develop a CNN for Pharma Image Segmentation.

Speaker: Anand Kadumberi, Data Scientist

Pre-requisites:
Elementary proficiency in python.
Familiarity with neural networks and Keras is desirable.

Hands-on Session
* TensorFlow docker container
* Kaggle (https://www.kaggle.com/c/data-science-bowl-2018)
Instructions for setting up and links to the Data
* We will use google colab for the hands-on session
https://colab.research.google.com/notebooks/welcome.ipynb

The only pre-requisite to run the same will be to copy the data and the python notebook required for the session, into every individual’s respective google drives and the session will be run on their google drives.

Materials to refer:
* Stanford Lectures: https://www.youtube.com/watch?v=nDPWywWRIRo
* https://medium.com/nanonets/how-to-do-image-segmentation-using-deep-learning-c673cc5862ef
* Books: Deep Learning by Ian Goodfellow
* Deep learning in Python by Francois Chollet.