Lab overview deck
10 slides, built from the same data as the rest of this site. The PowerPoint copy is generated from this page, so the two never disagree.
AIMS
Artificial Intelligence & Multimodal Systems
A research group building AI systems that learn from images, video, text, speech and sensor data, and stay reliable under uncertainty.
ComputerVisionVisionLanguageModelsRAGConversationalMemoryAIAgentsTrustworthyAI
Scan for papers, projects and people
aimsresearchlab.com
- aimsresearchlab.com
- github.com/aimsresearchlab
- TEC 340, Hattiesburg, MS
- Directed by Dr. Rabab Abdelfattah
What we do
Five lines of work
16 researchers
20 papers
8 active projects
-
Computer Vision Seeing and interpreting the visual world. -
Language & Agents Models that reason, act, and recover from failure. -
Multimodal AI Text, image, audio and sensor data together. -
Trustworthy AI Evaluation that stays valid under pressure. -
Human-Centered AI Systems people can direct, inspect and correct.
Project
SEAM

#HumanCenteredAI#LLMEditing
Measuring how much typed user speech gets absorbed into edited artifacts at unmarked within-turn seams.
DemoCode
Scan for the project page
Project
4D Gaussian Streaming

#ComputerVision#GaussianSplatting
Which Gaussians you pick as anchors barely matters: cheap samplers match the default strategy at deployment budgets, so the standard 8,192-anchor choice over-spends compute.
Does It Matter Which Gaussians You Pick in 4D Gaussian Streaming?
arXiv
Scan for the project page
Project
Same Ranking, Different Winner

#ConversationalMemory#Evaluation
The scoring target a memory benchmark picks can flip which LLM wins, even when the overall ranking looks stable.
Same Ranking, Different Winner: How Scoring Targets Shape LLM Memory Benchmarks
DemoarXivCode
Scan for the project page
Projects
#AIAgents#TrustworthyAI
Outcome Monitors
Recovery affordances that let an agent notice and repair tool calls that failed silently.
Outcome Monitors: Recovery Affordances for Silent Tool Failures
arXiv
#RAG#Evaluation
Fixed RAG Compression
A fixed compression layer in a RAG pipeline collapses the reader-scaling trend that the uncompressed setup shows.
Fixed RAG Compression Collapses Measured Reader Scaling
arXivCode
#ConversationalMemory#Evaluation
LAPSE
Memory consolidation flattens the temporal shape of user facts: an ongoing activity is stored as a standing one.
Memory Consolidation Flattens the Temporal Shape of User Facts
arXivCode
#ComputerVision#FoundationModels
Thin-Object Segmentation
A cross-domain survey of power-line, crack, and retinal-vessel segmentation, from classical pipelines to promptable foundation models.
From Classical Pipelines to Promptable Foundation Models: A Cross-Domain Survey of Thin-Object Segmentation for Power Lines, Cracks, and Retinal Vessels
Paper
#ComputerVision#EdgeAI
UAV Smoke Detection
A lightweight CNN screens every frame on the drone and hands only uncertain ones to a stronger model, cutting processing time by up to 56.9%.
Bayesian Optimization-Aided Hybrid Deep Learning Model for Lightweight UAV-Based Smoke Detection
Paperposter
People
Direction
Dr. Rabab Abdelfattah
Director, AIMS Lab
School of Computing Sciences and Computer Engineering
Dr. Sarah Lee
Director, School of Computing Sciences and Computer Engineering
University of Southern Mississippi
People
The lab
PhD Students
Olanrewaju Muili
Siyan Luo
Stone Barnard
Ticauris Stokes
Undergraduates
Sugam Panthi
Muhaiminul Yeamin
Sakshyam Sigdel
Shrabya Bhattarai
Joshua Johnston
Nhoojah Maharjan
Bibas Kandel
Bishesta Bohara
Jahandad Irfan
Collaborating Students
Ashim Dahal
Rashika Karmacharya
Alumni
Akram Hossain
Spring 2026
Murad Hasan
Spring 2026
Ticauris Stokes
Spring 2026
Publications
Selected papers
From Classical Pipelines to Promptable Foundation Models: A Cross-Domain Survey of Thin-Object Segmentation for Power Lines, Cracks, and Retinal Vessels
A. Hossain, N. Maharjan, M. Hasan, R. Abdelfattah, M. Ezz-Eldin, X. Wang, M. M. Fouda, K. Abdelfattah
IEEE Internet of Things Journal · 2026
Same Ranking, Different Winner: How Scoring Targets Shape LLM Memory Benchmarks
S. Panthi, R. Abdelfattah
Findings of EMNLP 2026 · 2026
Gaussian Splatting for Real-Time Scene Reconstruction: Methods, Systems, and Trade-offs
A. Dahal, R. Abdelfattah, N. Rahimi
Pattern Recognition · 2026
A comprehensive survey on facial expression generation: From GANs to LLM-guided multimodal models
M. Hasan, R. Abdelfattah, K. Abdelfatah, M. Fouda, A. Sherif
ICT Express · 2026
Full list: aimsresearchlab.com/publications/
News
Recently
- Nov 10, 2026 Accepted in Pattern Recognition: "Gaussian Splatting for Real-Time Scene Reconstruction: Methods, Systems, and Trade-offs"
- Nov 6, 2026 Congratulations to Sugam Panthi on receiving an Eagle SPUR undergraduate research grant from USM
- Oct 28, 2026 Congratulations to Sugam Panthi on receiving a $5,000 research grant in API credits from Pangram
- Sep 29, 2026 New preprint: "Memory Consolidation Flattens the Temporal Shape of User Facts" (LAPSE)
- Sep 25, 2026 S. Brandon Barnard joins the AIMS Lab to pursue his PhD studies
- aimsresearchlab.com
- github.com/aimsresearchlab
- TEC 340 · Hattiesburg, MS
AIMS
Artificial Intelligence & Multimodal Systems