A practical walkthrough from Classification network to Semantic Segmentation, let's do this one pixel at a time.

Abhishek Kumar (~vibrantabhi19)




Semantic Segmentation of an image is to assign each pixel in the input image a semantic class in order to get a pixel-wise dense classification. While semantic segmentation/scene parsing has been a part of the computer vision community since late 2007, but much like other areas in computer vision, a major breakthrough came when fully convolutional neural networks were first used by 2014 Long et. al. to perform end-to-end segmentation of natural images. An introduction to classification network and how it can be seen/converted to a segmentation network i.e a fully-integrated segmentation workflow, allowing you to create image segmentation and analyze the output of a segmentation network.

I will walk through my experience and the problems faced when trying to make a state of the art industry-grade implementation. Some of the points I will try to cover.

  • An intuitive introduction and visualization of the Convolution Neural Netowrk (the lifeline of Computer Vision).
  • An explanation and small demo of Classification Network using PyTorch.
  • Covering the bridge of Classification and Segmentation.
  • A small demo of fully-integrated segmentation workflow, allowing you to create visualize and understand segmentation datasets and visualize the output of a segmentation network.
  • Importance of data and problems faced by me working on Industry projects.
  • Wrap up and project discussion.


Zeal to learn new things would be enough but basic knowledge of Python would be good to go but the following are always encouraged: - Basic Knowledge of algebra and Python Libraries such as Numpy - Basic knowledge of working with Neural Network (not a strict requirement)

Content URLs:

Personal Blog

A naive Medium Blog sharing light on classification network

I will be putting the presentation links soon.

Speaker Info:

I am presently working as Deep Learning Scientist at Predible Health, here, we have build state of the art segmentation network for liver, tumour and vessel segmentations. I have spoken previously at Shri Mata Vaishno Devi University at their SFD celebrations and I am also speaking at PyCon Hyderabad and at MuPy (Manipal Institute of Technology's annual Python Conference) later this year. I have been a constant contributor in the open source world and have been attending PyCon and other conferences every year. An athlete, a Real Madrid F.C follower and a part time stand-up comedian (good enough to make you laugh). See you at PyCon.

Speaker Links:

Personal Blog

Github Link

LinkedIn Profile

Twitter handle


Feel free to drop a review or message regarding the talk or ML/DL in general.

Section: Data Analysis and Visualization
Type: Talks
Target Audience: Intermediate
Last Updated:

No comments added so far.

Login to add a new comment.