Pranav Gupta
Multimodal AI • Vision+Language • Generative Models
Master's student (incoming) in CSE at the University of Michigan, Ann Arbor. I build multimodal systems and study natural-language supervision for vision tasks.
Multimodal AI • Vision+Language • Generative Models
Master's student (incoming) in CSE at the University of Michigan, Ann Arbor. I build multimodal systems and study natural-language supervision for vision tasks.
I will be starting my Master's in CSE at the University of Michigan, Ann Arbor. My research interests revolve around Multimodal AI and natural language supervision in vision tasks. I'm currently a Research Intern at Stanford's PanLab working on multimodal AI in neuroimaging, and I also intern at DREAM:Lab (IISc) on deep learning for edge accelerators.
Past stints include Samsung, Upthrust, and AarogyaAI on research and data science projects.
ICVGIP 2024
Human-like danger assessment from videos with LLM-based reasoning and failures on subtle cues.
IEEE CONECCT 2024
Label-ontology guided pretext tasks improve audio classification via semi-supervised learning.
AAIMB 2023
A camera-first pipeline for attendance with privacy-preserving data flows.
IEEE CCEM 2020
Graph-based proximity modeling to estimate contagion transfer risk in gatherings.
Predict the next best word by parsing a screenshot of a partially-filled Wordle.
Query your search history content semantically to jump to the right page.
Parse a bill image and auto-create Splitwise entries end-to-end.
Evolve CNN kernels and pooling to optimize MNIST classification.
One-vs-all and one-vs-one SVMs using FaceNet embeddings on LFW.
Clean PyTorch/TensorFlow implementations for core vision & audio papers.
