Graduate Course · Semester I, 2026–27

ES 666 · Computer Vision

A graduate elective at IIT Gandhinagar spanning classical and learning-based computer vision — from camera geometry and multi-view reconstruction to CNNs, transformers, and generative models.

Google ClassroomEmail Instructor

Teaching AssistantsArjun Badola · Anupam Sharma · Dikshit Hegde · Hement S Mohanty

ES 666Course Code
4Credits · L3 T0 P0
C1 · C2Lecture Slots
AB 11/101Classroom

Course Map

A single schematic of the whole course: the geometric-vision pipeline up top, the six parts flowing from classical, proof-complete methods (Parts I–IV) into learning-based vision (Parts V–VI), the ten applications threaded throughout, and the four recurring throughlines. Click any part to jump to its lectures in the schedule.

Course MapThe whole of ES666 on one schematic — Parts I–IV design the function (classical, proof-complete); Parts V–VI learn it.GEOMETRIC VISION — FORWARD PROJECT, INVERSE ESTIMATE3D worldscene point XImagesx = PXMatchesx ↔ x′GeometryH · F · E · PStructureX, camera poseprojectmatchestimaterecoverCLASSICAL — DESIGN THE FUNCTION · PARTS I–IV · HOURS 1–18Foundations — jump to schedulePART IH1–3 ↗Foundationsimage formation · the SVDleast squaresFiltering — jump to schedulePART IIH4–6 ↗Filteringconvolution · Fourier · smoothingFeatures — jump to schedulePART IIIH7–10 ↗Featuresedges · Canny / Harris · scale spaceSIFT / SURF / ORBGeometry, Motion & Light — jump to schedulePART IVH11–18 +18a ↗Geometry, Motion & Lightprojective geometry · homographyDLT / RANSAC · calibrationepipolar F/E · MVS & SfMoptical flow · photometric stereoshape-from-Xclassical first, learned secondLEARNING — LEARN IT · PARTS V–VI · HOURS 19–28Learning for Vision — jump to schedulePART VH19–23 ↗Learning for Visionthe MLP · generalization & optimizers · CNNs · the VAE · vision applicationsAdvanced Paradigms — jump to schedulePART VIH24–28 ↗Advanced ParadigmsGANs · transformers · diffusion · flowsrepresentation & self-supervised learningTEN APPLICATIONS — THREADED THROUGH THE COURSE (SOLVED CLASSICALLY, THEN LEARNED)image classificationsemantic segmentationobject detectiondepth estimationoptical-flow estimationhomography estimationimage restorationimage relightingneural radiance fieldsHDR imagingFour throughlines:GeometryLearningGenerationRepresentation
The twenty-eight hours at a glance — each part assumes the last; ten applications are solved first by hand, then learned. Tip: the six part cards (↗) link to their rows in the lecture schedule.

How the Topics Connect

Beyond their linear order, the topics form a web of dependencies: linear algebra underwrites geometry, filtering feeds features, features drive matching and motion, and the whole classical toolkit becomes the inductive bias the learned models inherit — with every strand eventually meeting at the ten vision applications. This relationship map summarizes how the twenty-eight hours’ topics build on and reuse one another; colour marks the course part, and highlighted links name the key idea.

How the Topics ConnectA directed relationship map — arrows point from a topic to what it feeds; labelled links name the key idea. Click any topic to open its lecture.I · FoundationsII · FilteringIII · FeaturesIV · GeometryV · LearningVI · Advancedspace = frequencylearned convolutionminimize a residualHarris = LKcoarse-to-finesolves Ax = 0CNNs in disguiseSVD & least squares — open lectureSVD & least squaresImage formation — open lectureImage formationConvolution & LTI — open lectureConvolution & LTIFourier — open lectureFourierGaussian pyramids — open lectureGaussian pyramidsGradients & edges — open lectureGradients & edgesStructure tensor — open lectureStructure tensorScale space — open lectureScale spaceSIFT / descriptors — open lectureSIFT / descriptorsHomography · DLT — open lectureHomography · DLTRANSAC — open lectureRANSACCalibration — open lectureCalibrationEpipolar F, E — open lectureEpipolar F, EMVS & SfM — open lectureMVS & SfMOptical flow — open lectureOptical flowShape from light / X — open lectureShape from light / XMLP — open lectureMLPOptimization & reg. — open lectureOptimization & reg.CNN — open lectureCNNVAE — open lectureVAEGANs — open lectureGANsTransformers — open lectureTransformersDiffusion — open lectureDiffusionFlows — open lectureFlowsRepr. / SSL — open lectureRepr. / SSLTen vision applications — open lectureTen vision applications
A directed topic map — arrows show which topic feeds which; the brighter labelled links are the course’s recurring threads. Every topic node links to its lecture, and all strands converge on the ten vision applications.

Overview

This course builds the fundamentals and principles of computer vision, then develops both classical and learning-based solutions across a wide range of applications. It begins with the mathematical groundwork and classical pipeline — image formation, features, and geometry — and moves into modern deep learning for vision, closing with advanced applications such as segmentation, depth estimation, neural radiance fields, and HDR imaging. Students from any discipline gain the foundational knowledge needed to build computer-vision models using OpenCV and PyTorch / Keras / TensorFlow.

Offered in Semester I of 2026–27 by Prof. Shanmuganathan Raman. Course materials, announcements, and assignments are managed through Google Classroom (join code dfrc4n7z).

Logistics

Academic Year & Term
2026–27 · Semester I
Instructor
Shanmuganathan Raman
Teaching Assistants
Arjun Badola · Anupam Sharma · Dikshit Hegde · Hement S Mohanty
Credits
L3 · T0 · P0 · C4
Classroom & Slots
AB 11/101 · slots C1 and C2
Google Classroom
Join the class · code dfrc4n7z
Prerequisites · UG
ES 112 Computing, MA 101, MA 102, or equivalent
Prerequisites · PG
None

Course Contents

Mathematical Basics

Linear AlgebraCalculusOptimization

Classical Computer Vision

Camera & Image FormationImage Analysis & ProcessingFeature DetectorsFeature Descriptors & MatchingShape from ShadingPhotometric StereoDepth from DefocusSingle-View Geometry & Camera MatrixTwo-View GeometryHomographyEpipolar GeometryFundamental MatrixShape from StereoStructure from MotionActive Depth SensorsOptical Flow

Learning for Computer Vision

Discriminative & Generative ModelsMulti-Layer Perceptron (MLP)Overfitting & UnderfittingRegularizationOptimizersConvolutional Neural Networks (CNN)TransformersVariational Autoencoders (VAE)Generative Adversarial Networks (GAN)Diffusion ModelsRecent Advances

Computer Vision Applications

Image ClassificationSemantic SegmentationObject DetectionHomography EstimationDepth EstimationOptical Flow EstimationImage RestorationImage RelightingNeural Radiance Fields (NeRF)HDR Imaging

Textbooks & References

Learning Outcomes

By the end of the course, students will understand the fundamentals and principles of computer vision, and will be able to design and develop both classical and learning-based solutions for a variety of computer-vision applications. The tools covered are intended to help students across disciplines who need the foundational knowledge to build computer-vision models using OpenCV and PyTorch / Keras / TensorFlow.

Tentative Lecture Plan

1
Weeks 1–3
Fundamentals of Mathematics & Image Processing
2
Weeks 4–9
Classical Computer Vision
3
Weeks 10–12
Learning for Computer Vision
4
Weeks 13–14
Advanced Computer Vision Applications

Lecture Schedule

The course runs as twenty-eight lecture hours across six parts — classical, proof-complete methods first (Parts I–IV), then learning-based vision (Parts V–VI), with ten vision applications solved first by hand and then learned. A single-view Shape-from-X supplement (Lecture 18a) extends Hour 18. Per-lecture slides appear in the Slides column as each topic is covered; the course frontmatter deck — roadmaps, notation and conventions for all twenty-eight hours — is available now.

HrLectureSlidesReferences
Part I · Foundations — Hours 1–3
1Introduction & image formationImage as a function; sampling, quantization & Nyquist; pinhole & perspective projection; lens & depth of field; the sensing pipeline — demosaicing, gamma, CRF.PDFSzeliski (2022); Hartley & Zisserman (2004)
2Linear algebra IFour subspaces; spectral theorem; SVD existence & Eckart–Young; rank, norms & conditioning; worked 2×2 SVD.Trefethen & Bau (1997); Szeliski (2022)
3Linear algebra IINormal equations & residual orthogonality; projection matrix; pseudo-inverse via SVD; homogeneous Ax = 0; minimum-norm & total least squares.Trefethen & Bau (1997); Hartley & Zisserman (2004)
Part II · Filtering — Hours 4–6
4Convolution & LTI systemsSifting; LTI ⇒ convolution; FIR/IIR; BIBO stability; correlation & NCC; 2D separability; Toeplitz/circulant view.Oppenheim & Schafer; Szeliski (2022)
5Fourier analysisComplex exponentials as eigenfunctions; CTFT/DTFT/DFT/FFT; the convolution theorem; circular convolution; Gaussian self-duality; sampling & aliasing.Bracewell; Oppenheim & Schafer
6Image filteringBox & Gaussian filters; separability & integral images; derivative-of-Gaussian; median / bilateral / NL-means; anti-aliased pyramids; sharpening.Gonzalez & Woods (2018); Szeliski (2022)
Part III · Features — Hours 7–10
7Gradients, edges, LoG & DoGFinite differences; Sobel separability; LoG closed form; heat equation; DoG ≈ LoG; structure tensor; Hough transform.Marr & Hildreth (1980); Szeliski (2022)
8Canny & HarrisCanny optimal detector, NMS & hysteresis; auto-correlation → structure tensor; R = det M − k·tr²M; rotation invariance.Canny (1986); Harris & Stephens (1988)
9Scale space & blobsScale-space axioms; diffusion-equation proof; Gaussian/Laplacian pyramids; characteristic scale; DoG scale-space extrema.Lindeberg (1994)
10SIFT / SURF / ORBDoG keypoints; sub-pixel & edge rejection; orientation; 128-D descriptor; the ratio test; SURF; FAST/BRIEF/ORB; RANSAC count.Lowe (2004)
Part IV · Geometry, Motion & Light — Hours 11–18 (+ 18a)
11Projective geometry of the planeHomogeneous points & lines; incidence, join & meet; the line at infinity; conics; the cross-ratio.Hartley & Zisserman (2004), Ch. 2
12Transformations & the homographyProjective transformations; the homography (8 DOF); the transformation hierarchy.Hartley & Zisserman (2004)
13DLT & robust estimationDirect Linear Transform; Hartley normalization; robust fitting with RANSAC.Hartley & Zisserman (2004), §4; Szeliski (2022)
14Camera models & calibrationP = K[R | t]; intrinsics & extrinsics; camera centre; calibration; image of the absolute conic.Hartley & Zisserman (2004), Ch. 6
15Epipolar geometry: F & EThe epipolar constraint; essential & fundamental matrices; epipoles & epipolar lines.Hartley & Zisserman (2004), Ch. 9
16Multi-view stereo & SfMTriangulation; multi-view stereo; structure from motion; scene reconstruction.Szeliski (2022); Hartley & Zisserman (2004)
17Optical flow + synthesisBrightness-constancy (OFCE); Lucas–Kanade; coarse-to-fine flow; course synthesis.Lucas & Kanade (1981); Szeliski (2022)
18Photometric stereo & shapePhotometric stereo; shape from light; inverse rendering.Forsyth & Ponce (2012); Szeliski (2022)
18aShape-from-X supplementA companion to Hour 18 — four single-view shape cues: depth from focus; depth from defocus (coded apertures, dual pixels, deep optics); shape from interreflections (direct–global separation); shape from polarization (normals from angle & degree of polarization).Nayar–Ikeuchi–Kanade (1991); Levin et al. (2007); Atkinson & Hancock (2006); Kadambi et al. (2015)
Part V · Learning for Vision — Hours 19–23
19Neural nets & the MLPDiscriminative vs. generative models; the perceptron & the XOR problem; activations; universal approximation; backpropagation; SGD, initialization & vanishing gradients; nets as distribution transformers.Prince (2023); Bishop & Bishop (2024)
20Generalization, regularization & optimizersBias–variance & the over/under-fitting curve; capacity & double descent; train/val/test; L2/L1, dropout, early stopping, augmentation; batch & layer norm; momentum, RMSProp, Adam; LR schedules.Prince (2023); Goodfellow, Bengio & Courville (2016)
21Convolutional neural networksLocality, weight sharing & translation equivariance; receptive fields; pooling; LeNet → VGG → ResNet; transfer learning; CNNs as image-to-image; the ViT bridge.Torralba, Isola & Freeman (2024); Prince (2023)
22Generative models & the VAELatent-variable generators g(z, y); autoencoders; variational inference & the ELBO; the reparameterization trick; the VAE & β-VAE; density vs. energy models; a map of GANs, flows & diffusion.Torralba, Isola & Freeman (2024); Bishop & Bishop (2024)
23Computer-vision applicationsOne backbone, many heads — classification, semantic segmentation (FCN/U-Net), detection (Faster R-CNN/YOLO); homography, depth & optical-flow estimation; restoration & super-resolution; relighting; NeRF; HDR.Torralba, Isola & Freeman (2024); Szeliski (2022)
Part VI · Advanced Paradigms — Hours 24–28
24Generative Adversarial NetworksThe adversarial game; minimax & the optimal discriminator → JS divergence; the non-saturating loss; mode collapse; WGAN; DCGAN, cGAN, pix2pix, CycleGAN, SRGAN, StyleGAN; FID.Goodfellow et al. (2014); Torralba, Isola & Freeman (2024)
25TransformersTokens & attention; scaled dot-product self-attention; multi-head attention; the transformer block; ViT — “CNNs in disguise”; positional encodings; DETR, Swin, SegFormer, DPT.Torralba, Isola & Freeman (2024)
26Diffusion modelsForward noising; predicting the added noise; reverse sampling (DDPM/DDIM); U-Net & time embedding; classifier-free guidance; latent diffusion; cross-attention text-to-image; ControlNet.Ho et al. (2020); Torralba, Isola & Freeman (2024)
27Flow modelsExact likelihood; change of variables & the log-det Jacobian; coupling layers (RealNVP); Glow 1×1 convolutions; continuous flows (Neural ODE); flow matching; SRFlow, compression, anomalies.Torralba, Isola & Freeman (2024)
28Representation & self-supervised learningEncoders & embeddings; autoencoders; pretext tasks; contrastive InfoNCE; alignment & uniformity; SimCLR, MoCo, BYOL; MAE; CLIP, zero-shot & open-vocabulary vision.Torralba, Isola & Freeman (2024)

Grading Scheme

Assignments40%Exam 230%Exam 120%Attendance (random)10%
Assessment weights — Assignments 40%, Exam 2 30%, Exam 1 20%, and random attendance 10% (total 100%).

Honor Code

Academic integrity. Cheating of any form in assignments and examinations is treated as a severe violation of the Institute’s honor code. Any student found in violation will be reported immediately to the SSAC.