We love teaching. That is why we hold internships and schools, organize applied computer vision meetups, and deliver open lectures at universities. Until autumn 2020, our audience mainly included students, interns, and developers, but now it also has Kharkiv university teachers on the list. Being invited by Kharkiv IT cluster to deliver a course on computer vision and machine learning for lecturers of IT specialties working at Kharkiv universities and colleges who seek to improve their qualifications, we joined the Prof2IT project and contributed to educators’ professional growth.
What is Prof2IT?
The Prof2IT project, held jointly by Kharkiv IT Cluster and Kharkiv University of Technology “Step”, aims at helping Kharkiv university teachers of IT specialties improve their skills and update their knowledge on specific IT and data science disciplines. Additionally, it is targeted to update the course materials they teach in accordance with the modern requirements of the IT industry.
With this project brought to life, Kharkiv IT Cluster contributed to direct communication between the IT industry and the teaching community. IT businesses are interested in ensuring that people who graduate from universities and come to work for them are highly competent and the market is saturated with highly-qualified young talents. Is it always the case? Unfortunately, no. How can we help as a company? We can train a teacher, and let the geometric progression of spreading knowledge do the rest.
Course on Computer Vision and Machine Learning
Our online training course “Applied Computer Vision: from Classic Image Processing to Machine Learning and Deep Learning” started on November 5, 2020, and lasted for two months. Twenty lecturers from ten different Kharkiv educational institutions including but not limited to Kharkiv National Polytechnic University “NTU KhPI”, Kharkiv National University of Radioelectronics and V. N. Karazin Kharkiv National University joined the course. All the materials were delivered by our experienced team:
What did the course include? Seven lectures divided into two modules and 2 workshops that covered all the basics both in classical computer vision algorithms, machine learning, and deep learning. The listeners could update their knowledge on the theoretical foundations of the formation of digital images and their analysis using different tools.
In the first part, we have covered some basics in image understanding and processing like color spaces, histograms, filtering, image gradients, thresholding, and morphological operations. The classical CV module also included feature extraction and matching; lecture on detection and tracking which included an overview of homography, RANSAC, camera calibration, optical flow and template-based tracking, and finally lecture on feature crafting, the bridge between classical approaches and machine learning.
The second module featured an introduction to both classical machine learning algorithms and neural networks. We covered the very basics of deep learning (neuron model, backpropagation, gradient descent, etc.) as well as the evolution of neural network architectures, different types of convolutional layers and regularization techniques, GANs, autoencoders and different task-specific architectures of neural networks.
With this course, we hope to contribute not just to the professional development of the teachers who attended the course but also to the modernization of the corresponding training courses in universities and colleges of the city.