v2.11.0 (5648)

Cours scientifiques - DS-télécom-13 : Deep Learning for Computer Vision

Domaine > Mathématiques appliquées.

Descriptif

This course explores advanced topics in computer vision, focusing on deep learning techniques and recent developments in the field. Through a combination of lectures and practical sessions, students will develop a comprehensive understanding of key research challenges in computer vision and the corresponding methodologies.

effectifs minimal / maximal:

/120

Diplôme(s) concerné(s)

Pour les étudiants du diplôme M2 Data Science

Basic knowledge of deep learning is required.

Format des notes

Numérique sur 20

Littérale/grade réduit

Pour les étudiants du diplôme M2 Data Science

Le rattrapage est autorisé (Max entre les deux notes)
    L'UE est acquise si Note finale >= 10
    • Crédits ECTS acquis : 3 ECTS

    Programme détaillé

    Course Outline:

    Week 1-2: Introduction to the main vision tasks

    • Object detection and segmentation

    • Depth estimation

    • Human pose estimation

    • Action recognition

    • Practical sessions

    Week 3-4: Self-Supervised Learning, Few-Shot Learning and Domain Adaptation

    • Introduction to self-supervised learning

    • Contrastive learning

    • Few-shot learning: Problem formulation and techniques

    • Domain adaptation: Theory and applications

    • Recent advances in few-shot learning

    • Practical session

    Week 5: Recent Advances in Image and Video Generation

    • Recap on Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs) and diffusion models

    • Conditional image synthesis

    • Text-to-image generation

    • Video generation and editing

    Week 6-7: Deep Architectures for Point Cloud Perception

    • Introduction to point cloud data

    • PointNet and PointNet++

    • Point cloud segmentation

    • Object recognition in point clouds

    • Practical session

    Please note that the specific topics and content may be adjusted based on the needs and preferences of the students and the availability of resources.

    Veuillez patienter