Pysangamam 2018: Workshop Content
Computer Vision through the Ages
Sep 9, 2018 · 2 min read
Original Post here: https://iitmcvg.github.io/conferences/pysangamam-content/

Thank you for the wonderful participation yesterday at PySangamam 2018.
For those who missed the workshop, the following is the outline of topics we covered:
What is computer vision?
- Computer Vision through the ages
- Image and video representations in computers (pixels, colour spaces).
- Why is computer vision hard?
- Examples of different computer vision problems (detection, segmentation, shape from X, etc).
Computer Vision through the ages
- Gradient Based Methods (HOG, HAAR).
- Scale invariant Feature Matching Algorithms (SIFT and SURF).
- Other tasks with classical Computer vision.
- Issues with classical computer vision.
Rise of learning methods
- Youtube cats, training on CPUs.
- SAIL — the era of GPUs.
- In a nutshell, ML+ NN = DL.
- Alexnet and the impact of Imagenet.
The Deep Learning Era in flow
- Introduction to Machine Learning, benefits of end-to-end solutions.
- Using ML algorithms for Computer vision tasks (Hello world (MNIST with keras)).
- Feed forward neural networks for MNIST classification.
- Introduction to Convolutional Neural Networks (CNNs).
- Using CNNs for Image Recognition .
Concept of Transfer Learning
- Why transfer learn; Less data? Not always a problem.
- Concept of Bottlenecks.
- A hands on example.
- Practitioner advice; Closing comments
Workshop Content
As promised, here are our slides and repo’s used:
Feel free to write back to us at cvigroup[dot]cfi[at]gmail[dot]com, or follow us on facebook or twitter for more updates!
Slides
Update 9th September 2018: We have embedded the slides in this post below.
Note: We will soon be posting stories about the other 3 talks delivered at Pysangamam too.
