yandexdataschool/deep_vision_and_graphics

Course about deep learning for computer vision and graphics co-developed by YSDA and Skoltech.

What it solves

This repository hosts all materials for the "Deep Vision and Graphics" course taught at Yandex School of Data Analysis (YSDA) in fall 2025. It replaces the older "Deep Learning" course with a sharper focus on applying deep learning to computer vision tasks — so students get a structured, updated path from classic neural networks to modern generative and 3D vision techniques.

How it works

The repo is organised by weeks: each week* folder contains lecture and seminar materials, while homework* folders hold assignments. The syllabus moves from fundamental neural network concepts (backprop, convolutions, batches, augmentations) through advanced architectures (transformers, mixers), then into dense prediction (segmentation, super-resolution), object detection and pose estimation, representation learning, latent and generative models (GANs, diffusion), and finally shape/motion estimation and novel view synthesis (neural radiance fields, neural renderers). Administrative details live on the YSDA LMS for enrolled students, and any issues or contribution ideas are handled via the GitHub issue tracker.

Who it’s for

It is aimed at YSDA students taking the course, and at anyone who wants to follow a rigorous, up-to-date curriculum in deep learning for computer vision — especially those who already know basic deep learning and want to go deeper into visual applications.

Highlights

  • Complete weekly schedule covering 12 topics, from CNN basics to diffusion models and neural radiance fields
  • Includes both lectures and seminars, plus separate homework folders
  • Taught by experienced course staff with contributions from multiple instructors
  • Successor to the well-known Practical Deep Learning course, modernised for current CV research
  • Open to community contributions and bug reports via GitHub issues

Related

  • Project
    ZZZHANG-jx/Recommendations-Document-Image-Processing

    This repository is a carefully organised collection of academic papers, datasets, and benchmark results for document image processing, covering tasks such as shadow removal, dewarping, deblurring, and appearance enhancement. It serves as a reference hub for researchers and developers looking to understand or apply state-of-the-art methods to clean up photographed documents.

  • Project
    Kohulan/DECIMER-Image_Transformer

    DECIMER is an open-source deep learning tool that automatically recognizes chemical structures in images and converts them into SMILES strings, the standard text format for molecules. It helps chemists digitize printed or hand-drawn chemical structures without manual data entry.

  • Project
    fregu856/papers

    A personal, public collection of over 550 research papers that the author has read since 2018, each with detailed comments, annotated PDFs, and organized into categories across machine learning and healthcare topics.

  • Project
    ilastik/ilastik

    An interactive machine learning toolkit for segmenting, classifying, tracking, and counting cells and other experimental image data.