Computational Perception     COMP 546  
Winter 2017
Tues/Thurs  11:35-12:55
 BURN 1B 23



Instructor:    Professor Michael Langer
                      School of Computer Science
Office:           ENGMC 329
Tel:               514-398-3740
Email:           langer  [at] cim.mcgill.ca
Office Hours:  Tues and Thurs 1-2 or by appointment
Teaching Assistant (T.A.)   David Bourque
Email:             david.bourque [@] mail.mcgill.ca
Office:     ENGMC 337
Office Hours:   by appointment
Announcements
  • Assignment 4 (PDF)   (I.raw)

  • Assignment 3 (PDF)   (solutions)
    FEEDBACK: I printed out hard copy of each assignment and made notes on it.   You can pick yours up from my office anytime.   Or if you can't make it,  let me know and I will scan and email the comments to you.

  • Assignment 2 (PDF)   (code)   (solutions)

Resources

LECTURE SCHEDULE
  1. introduction (slides) (notes)
    intro, course outline, origin of eyes and spatial vision
VISUAL IMAGE FORMATION
  1. geometry (slides)   (notes)
    visual angle, aperture, image projection, binocular disparity, sampling, thin lens equation
  2. blur, photometry (slides)   (notes)
    depth of field, accommodation, aging and abnormal vision; shading, shadows, highlights, radiance
  3. photoreceptors, color (slides)   (notes)
    spectra: emission, reflectance, absorptance; rods and cones, metamers, color displays, color blindness
EARLY VISION PATHWAY
  1. retina (slides)   (notes)
    spikes, color opponency, center-surround DOGs, cross-correlation
  2. orientation selectivity in V1 (slides)     (notes)
    simple and complex cells, Gabor models
  3. retinotopic maps, binocularity in V1 (slides)   (notes)  
    retinal receptive field size and eccentricity, orientation columns, ocular dominance columns, binocular complex cells
  4. image motion (slides)   (notes)  
    XYT, motion constraint equation, aperture problem, intersection of constraints (IOC), 3D Gabors, sketch of MT
3D SURFACE PERCEPTION
  1. depth from blur, binocular steropsis (slides)   (notes)  
    blur and occlusions, slanted planes, tilt-shift illusion; Panum's fusional area, random dot stereograms, disparity space, accommodation-vergence conflict
  2. egomotion and depth from parallax, eye movements (slides)   (notes)  
    translation and rotation components; VOR, smooth pursuit, saccades
  3. shape from texture and shading (slides)   (notes)  
    slant & tilt; texture size, density & foreshortening cues, diffuse vs. specular reflections
MEASURING AND MODELLING PERFORMANCE
  1. psychophysics (slides)   (notes)  
    psychometric curves, thresholds, contrast and disparity sensitivity
  2. maximum likelihood method (slides)   (notes)
    examples of likelihoods, probability review
  3. cue combinations, Bayesian models (slides)   (notes)
    priors, MAP, depth reversal ambiguity
LINEAR SYSTEMS THEORY & IMAGE FILTERING
The material in lectures 14-16 gives another view of 'feature maps'. Its main purpose, though, will be to provide the mathematical foundations for the audition lectures.
  1. convolution (slides)   (notes)
    impulse response functions
  2. Fourier transform (slides)   (notes)
    examples, convolution theorem, inverse Fourier transform
  3. filtering (slides)   (notes)
    white noise, low/band/high pass filters, Gaussian and Gabor, 2D Fourier transforms
'BOTTOM UP' VERSUS 'TOP DOWN' PROCESSING  
  1. attention (slides)   (notes)
    feature maps, saliency, visual search
  2. perceptual organization, object recognition (slides)     (notes)
    Gestalt laws of grouping, mid-level vision, recognition: from Marr to machine learning
AUDITORY IMAGE FORMATION  
  1. sound 1 (slides)   (notes)
    waves, intensity, dB, interaural differences
  2. sound 2 (slides)     (notes)
    music and speech sounds, spectrograms
AUDITORY SYSTEM & SPATIAL HEARING        
  1. head and ear (slides)   (notes)  
    head and outer ear (HRIR, HRTF), inner ear and neural coding, critical bands
  2. auditory pathway in brain, source localization (slides)   (notes)  
    from brain stem to cortex (A1), duplex theory, level and timing differences
  3. echolocation and recognition by bats and porpoises (slides)   (notes)  
    constant frequency, frequency modulation, echos

Exercises, Assignments, Exams






  Exercises 1


  Exercises 2


  Exercises 3


  Exercises 4       Assignment 1 (PDF)   (starter code)   (solutions)

  Exercises 5


  Exercises 6


  Exercises 7         Midterm 1 with solutions (PDF)



  Exercises 8


  Exercises 9


  Exercises 10


  Exercises 11


  Exercises 12


  Exercises 13

  midterm 2 (with solutions and grading scheme) (PDF)
  posting for midterm 2 bonus (slight rewording of some questions) (PDF)


  Exercises 14


  Exercises 15


  Exercises 16


  Exercises 17 and 18 (TODO - not sure if I will have time)




  Exercises 19


  Exercises 20



  Exercises 21

  Exercises 22

  Exercises 23