Assignment #4: Two grayscale faces (Ferdowsipoor and Modiri) are attached below. They have the same size (250 x 360 pixels). You write a code (preferably in Matlab or Python) to morph them. You generate 10 faces in the morphing continuum (100% F, 90% F, ..., 50% MF, ..., 90% M, 100% M). The 50% MF is an average face. Since the facial features of these two faces are very different, morphing would not be an easy task; a simple pixel-based weighted averaging would not work at all (Hossein and I already tested this). Given the fact that this is a challenging assignment, you have two weeks to work on it. If you couldn’t implement a code, don’t worry - you can use a software to generate morphed faces. Obviously if you do it using your own code, you get extra points.

Whenever you are done, please send your code, denoscription of the algorithm used, and 10 morphed faces to our fabulous TA, Hossein Rafipoor (h.rafipoor@gmail.com).

Note: to make morphing easier, you may extract facial features manually.

Tomorrow (session #5), I will continue the topic of face perception. I will also talk about the principles of shape coding in the brain.

You may find this page useful and interesting:

http://ohzawa-lab.bpe.es.osaka-u.ac.jp/ohzawa-lab/teaching/AA_RFtutorial.html

Space-Time Receptive Fields of Visual Neurons

This is an interesting paper:

Slides of session 5 and session 6 👇

Hi everyone, please complete the assignment #4 (morphing of faces) by Tuesday this week.

Assignment #5: Design a simple experiment to demonstrate blind spot in retina. Also explain why we don’t perceive blind spot in our natural vision.

Amin would have an additional assignment to tell us the theories proposed for explaining Hermann grid illusion :)

In the first session after Eid (Yekshanbe, 19 Farvardin), Amir Farzmahdi (a brilliant PhD student of cognitive neuroscience in IPM) will be an invited lecturer to speak about Deep Neural Networks and Human Vision. This would be an interesting session, and it is highly recommended.

Slides of sessions 7,8,9 👇

Spike-triggered average is also called “reverse correlation”.

Today is 4shanbe-soori! Participation in today’s class is optional. We will show a movie about recording of V1 cells.

A movie that was shown in session 10: