460-4107/02 – Image Analysis II (ANO II)
| Gurantor department | Department of Computer Science | Credits | 4 |
| Subject guarantor | Ing. Radovan Fusek, Ph.D. | Subject version guarantor | Ing. Radovan Fusek, Ph.D. |
| Study level | undergraduate or graduate | Requirement | Optional |
| Year | 2 | Semester | winter |
| | Study language | English |
| Year of introduction | 2015/2016 | Year of cancellation | |
| Intended for the faculties | FEI | Intended for study types | Follow-up Master |
Subject aims expressed by acquired skills and competences
The goal of the course is to get the student acquainted with modern methods of image analysis that can be used in the area of object detection and recognition. An integral part is also application of this methods in the real world (e.g. detection and recognition of faces, localization of pedestrians, detection of cars).
Teaching methods
Lectures
Tutorials
Teaching by an expert (lecture or tutorial)
Summary
The following topics are covered: Modern methods of object detection and object recognition. Typically, the approaches are based on the image descriptors that are combined with the machine learning methods. The principles and aplications of deep learning and convolutional neural networks are also covered (detection of vehicles, pedestrians, faces).
Compulsory literature:
1. Chollet, F.: Deep Learning with Python. Manning, ISBN-13: 978-1617294433, 2017
2. Gonzalez, R. C., Woods, R. E.: Digital image processing, New York, NY: Pearson, ISBN-13: 978-0133356724, 2018
3. Zhang, A., Lipton, Z.C., Li, M., Smola, A.J.: Dive into Deep Learning, https://d2l.ai, 2020
Recommended literature:
1. Burger, W., Burge, M., J.: Principles of Digital Image Processing: Fundamental Techniques, Springer, ISBN-10: 1848001908, ISBN-13: 978-1848001909, 2011
2. Brahmbhatt, S.: Practical OpenCV (Technology in Action), Apress, ISBN-10: 1430260793, ISBN-13: 978-1430260790, 2013
3. Gary Bradski, Adrian Kaehler: Learning OpenCV 3: Computer Vision in C++ with the OpenCV Library, O'Reilly Media, 2017
Additional study materials
Way of continuous check of knowledge in the course of semester
Conditions for granting the credit:
The tasks that form the program of exercises must be worked out.
Exam - oral examination.
E-learning
Materials are available on the educator's website:
https://mrl.cs.vsb.cz//people/fusek/ano2_course.html
Other requirements
No further requirements are imposed on student.
Prerequisities
Co-requisities
Subject has no co-requisities.
Subject syllabus:
Lectures:
* Basic concepts of object detection in images. Methods for face detection in images. Haar features (AdaBoost, Viola-Jones). Local Binary Patterns (LBP), Histograms of Oriented Gradients (HOG) and their use for object analysis. Methods for pedestrian detection in images.
* Convolutional neural networks (basic principles, description of layers). Modern variants of convolutional neural networks (e.g. VGGNet, GoogLeNet, ResNet).
* Description of convolutional networks for object localization (e.g. R-CNN, Faster R-CNN, YOLO, SSD).
* Description of generative networks (e.g. DCGAN, Diffusion-GAN).
* Transformer networks (especially Vision Transformer – ViT) and their use in image analysis.
* Convolutional neural networks for image segmentation (encoder-decoder networks, U-Net).
* Keypoint detectors and descriptors (e.g. SIFT, SURF).
* Optical systems in the area of self-driving vehicles, IR image processing, LIDAR image processing, depth image analysis, use of depth sensors (RealSense, Kinect).
Computer Labs:
* Development of the detector for the selected object of interest, implementation of the sliding window method, preparation of data for the training and testing phases of the detector.
* Detection based on Haar-type features, detection using local binary patterns, analysis of objects using gradients (HOG method).
* Experiments with convolutional neural networks, exploring the parameters of convolutional networks.
* Application of different types of convolutional networks (e.g. VGGNet, GoogLeNet, ResNet), comparison of detectors.
* Practical use of localization methods based on convolutional neural networks (e.g., R-CNN, Faster R-CNN, YOLO).
* Practical use of generative networks for data augmentation (DCGAN, Diffusion-GAN).
* Experiments with image segmentation using encoder-decoder networks (U-Net).
* Practical use of transformer networks for object analysis in images.
* Analysis of objects in IR and depth images (RealSense, Kinect).
Conditions for subject completion
Occurrence in study plans
Occurrence in special blocks
Assessment of instruction
Předmět neobsahuje žádné hodnocení.