Welcome to 7th ADASIVA 2025 Summer School !!!

About

The 7th Summer School on Advances in Deep Architectures for Signal, Image and Vision Applications (ADASIVA 2025) is being organized by Computer Vision and Biometrics Lab, Indian Insttiute of Information Technology, Allahabad during July 07-12, 2025 in virtual mode. ADASIVA 2025 is an advanced training program designed to provide participants with a comprehensive understanding of the latest trends and techniques in deep learning as applied to signal, image, and vision domains. Taking place in a collaborative and interdisciplinary environment, the school brings together leading researchers and practitioners to deliver a rich mix of theoretical lectures and practical sessions. Topics covered include state-of-the-art deep architectures such as convolutional neural networks (CNNs), vision transformers, generative models, and diffusion model, with a focus on their applications in fields like computer vision, biometrics, medical imaging, remote sensing, and multimedia analysis.

ADASIVA 2025 is aimed at students, early-stage researchers, and professionals from academia and industry who are eager to deepen their expertise in AI-driven visual and signal processing. The program emphasizes cutting-edge developments in deep learning, hands-on learning through coding labs and project-based challenges, and opportunities for mentoring with experts. Beyond technical content, the summer school encourages networking, idea exchange, and collaboration across disciplines, fostering a vibrant community engaged in pushing the frontiers of deep learning research and its real-world deployment.

Objectives: The key objectives of the 7th Summer School on Advances in Deep Architectures for Signal, Image and Vision Applications (ADASIVA 2025) are:

• To provide in-depth knowledge of advanced deep learning architectures such as CNNs, transformers, generative models, and diffusion models, with a focus on their applications in signal, image, and vision processing.
• To bridge theory and practice by offering hands-on sessions and coding labs that enable participants to implement and experiment with state-of-the-art models.
• To foster interdisciplinary collaboration and knowledge exchange among participants from diverse fields such as computer vision, signal processing, biomedical engineering, and biometrics.
• To expose participants to emerging research trends and challenges in deep learning for vision and signal applications through expert talks, case studies, and interactive discussions with leading experts.
• To support the development of innovative ideas and solutions by engaging participants in projects, and mentoring sessions that encourage creativity and problem-solving in real-world contexts.

Topics to be Covered (Tentative):
Following are the key topics to be covered at the 7th Summer School on Advances in Deep Architectures for Signal, Image and Vision Applications (ADASIVA 2025):

• Fundamentals of Deep Learning: Neural networks, backpropagation, optimization techniques, Regularization, generalization, and training strategies.
• Convolutional Neural Networks (CNNs) and Variants: Architectures for image classification, object detection, and segmentation, Lightweight and efficient CNN models for edge deployment.
• Adversarial Learning: Adversarial Attack, DeepFake, Generative Adversarial Network (GAN) and variants for image generation, image-to-image translation and applications.
• Recurrent Neural Networks (RNNs) and Variants: Architectures for sequential and time-series data, RNN, LSTM and GRU.
• Attention Mechanisms and Vision Transformers: Self-attention in vision tasks, transformers and vision transformers.
• Diffusion Architectures: Denoising Diffusion Probabilistic Models (DDPM), forward process, reverse process, Score-based Diffusion Model, Latent Diffusion Model.
• Explainability in Deep Learning: Techniques for Visualization and Explainable AI, Ethical Considerations in XAI.
• Multimodal Learning and Fusion Techniques: Integrating image, text, and signal data, Cross-modal representation learning.
• Real-World Applications and Case Studies: Computer Vision, Biometrics, Medical imaging, etc.

Each topic will be explored through expert lectures, practical labs, and interactive discussions to provide a well-rounded understanding of both foundational concepts and emerging advances.