This course offers a comprehensive journey through the field of computer vision, bridging classical techniques with modern deep learning approaches. Students begin with foundational concepts such as camera calibration, stereo vision, homography, and projective geometry, before progressing to feature extraction methods like SIFT, Harris Corner Detection, and Optical Flow. The curriculum then transitions into deep learning, covering Convolutional Neural Networks, transfer learning, object detection architectures such as R-CNN and YOLO, and sequential models including RNNs and Transformers. Advanced topics include Vision Transformers (DETR, SETR, SWIN), self-supervised learning, and Vision-Language Models (VLMs), reflecting the cutting edge of the field. Assessment is project-driven and application-focused, with a major project and final exam together comprising 80% of the grade, complemented by a student presentation, ensuring that learners develop both theoretical understanding and practical problem-solving skills
Week Topic
1- Introduction
2- Camera Fundamentals & Calibration
3- Simple Stereo, Homography
4- Projective Geometry, Essential Matrix, Fundamental Matrix,Triangulation
5- Features, SIFT, Harris Corner Detector,Optical Flow
5- Deep Learning Fundamentals
6- CNNs, Transfer Learning
7- Object Detection, R-CNN, YOLO
8- Sequential Models, Image Captioning, Video Analysis
9- Transformer Networks
10- Transformers in Computer Vision,DETR SETR,SWIN Transformers
11- Self supervised Learning & VLMs
12- Student Presentations
13- Student Presentations
14- Student Presentations
Grading
%40 Project
%40 Final
%20 Presentation