Shehreen Azad

I am a fourth year PhD student at Center for Research in Computer Vision (CRCV), University of Central Florida (UCF), under the supervision of Dr Yogesh Singh Rawat.

I have a broad interest in deep learning and computer vision. My research mainly focuses on long-form video understanding and multimodal reasoning.

Currently looking for internship positions for Summer'26! Feel free to drop me an email.

Email  /  Google Scholar  /  Resume  /  LinkedIn  / 

Updates

  Aug'25: Broadening Participation Award to attend ICCV 2025.
  Jun'25: DisenQ (Fist-author paper) got accepted in ICCV 2025 as Highlight. ⭐️🔥
  Mar'25: Depth and Height Perception Benchmark (First-author paper) got accepted in CVPR MMFM Workshop 2025.
  Feb'25: HierarQ (First-author paper) got accepted in CVPR'25. 🔥
  Apr'24: Diversity, Equity and Inclusion Award to attend CVPR 2024.
  Mar'24: Distribution Shift Benchmark got accepted in CVPR MMFM Workshop 2024.
  Mar'24: Compositional Reasoning Benchmark got accepted in CVPR MMFM Workshop 2024.
  Feb'24: Activity-Biometrics (First-author paper) got accepted in CVPR'24. 🔥

Publications

Below is a selected list of my works (in chronological order), representative papers are highlighted.

Developing a novel memory-augmented framework to equip Multimodal Large Language Models (MLLMs) with real-time, continuous understanding of long-form streaming video.

A novel disentanglement-based Multimodal Large Language Model (MLLM) based architecture for robust activity-aware person identification.

The first diagnostic benchmark for specifically evaluating the depth and height perception capabilities of Multimodal Large Language Models (MLLMs).

A novel task-aware and hierarchical framework to equip Multimodal Large Language Models (MLLMs) for efficient arbitrarity long-form video understanding.

Investigated the robustness of Multimodal Foundation Models (MFMs) in the face of distribution shift caused by perturbations and corruptions across 17 different categories and 5 different severity levels.

Investigated and improved the relational, compositional and contextual understanding of Multimodal Large Language Models (MLLMs) through 3 novel benchmarks.

Proposed the novel task of activity-biometrics, which scales traditional gait-based person reID methods to activity-aware person reID. Developed a novel disentanglement based framework to improve robust activity-aware person reID.

Awards / Recognitions

Professional Service

This website has been adopted from this source code.