Basic Information
- Project Title / Domain: Vision-Language Models for Whole Slide Image Classification
- Name: Shounak Das
- Guide: Amit Sethi
- Project Type: implementation-driven
Short Description of Project
Working on vision-language models for whole slide images, including multi-resolution pipelines, prompt tuning with CLIP/LLaVA, and improving alignment for better zero-shot performance
Whom did you work with?
PG students, The prof directly
Tools / Simulation / Software / Hardware
PyTorch, Docker, remote GPU servers for training and experimentation
Expectations from Guide
Expectations for 8th Sem & Summer
No, did an internship. Had a few meets with my guide and a PhD student
Load: 8th Sem vs 9th (Placement) Sem
In 9th sem, the prof allowed slightly less load in November after the DDP1 presentation for placements
Summer on Campus
Didn’t stay on campus, worked during internship with occasional discussions, so moderate progress
Is DDP Guide Same as SRE/RnD Guide?
Yes
How did your SRE help with your DDP?
Provided initial exposure to the problem and relevant literature about WSIs
End Deliverables
A working pipeline, experiments and analysis and a novel contribution