Dr. Anshul Ujlayan

Intelligent Vision and Computing: Bridging Algorithmic Frontiers with Real-World Impact

Abstract:

The rapid convergence of deep learning, large-scale self-supervised pre-training, and accelerated hardware has elevated computer vision from a narrow pattern-recognition discipline into a foundational pillar of modern artificial intelligence. This keynote presents a comprehensive and practice-oriented survey of Intelligent Vision and Computing spanning convolutional architectures, Vision Transformers, multimodal foundation models, and generative AI systems with an emphasis on practical implementation for researchers and graduate students. The discussion will help to trace the algorithmic evolution from AlexNet (2012) through the Vision Transformer paradigm (2020) to today's large vision-language models such as GPT-4o, Gemini 2.0, and open-source counterparts including LLaVA-NeXT and InternVL2. The high-impact application domains are examined in depth, including medical image analysis, autonomous vehicles, robotics, environmental monitoring, and precision agriculture. We further provide a structured, 90-day roadmap enabling students and early-stage researchers to move from foundational Python skills to fine-tuning state-of-the-art foundation models and publishing reproducible results. Ethical considerations algorithmic bias, privacy, energy sustainability, and adversarial robustness  are treated as first-class research concerns rather than peripheral footnotes. The keynote concludes by identifying seven open research challenges that represent high-impact opportunities for the next generation of scholars in this field.