My Projects

Conditional Prompt Learning for Vision-Language Models

Vision-Language Models like CLIP can match images and text well and support zero-shot learning. However, they often struggle with new classes and domains, especially when labeled data is limited. Methods like CoOp use fixed prompts but fail to adapt to different image types. This project aims to improve generalization by learning prompts that change based on the image's features, helping the model perform better in unseen scenarios.

About Program

The PG Level Advanced Certification Course in Deep Learning is offered by TalentSprint in collaboration with Center of Continuing Education at IISc is a 10-month weekend program that enables both aspiring and practicing AI/ML professionals to build expertise in Deep Learning. The program covers the essential theoretical foundations of deep learning and teaches students how to apply them in the real world effectively. The course is best suited for individuals with programming knowledge who want to create a practical understanding of how machine learning algorithms can be developed and optimized for hardware. For more details visit https://iisc.talentsprint.com/deeplearning/index.html