Welcome to the AVIS Lab
The Advanced Vision & Intelligent Systems (AVIS) Lab at the Department of Robotics and Mechatronics Engineering, University of Dhaka, builds AI that can see, reason, and act in the physical world. Led by Dr. Md Mehedi Hasan, we work where computer vision, machine learning, and robotics meet. Our drones navigate with vision-language models. Our agents learn long-horizon behaviour from offline data. Our models diagnose crop and human disease and explain their reasoning. We publish at venues including Findings of ACL, PLOS One, Plant Methods, and IEEE Access, and partner with agricultural and medical institutions across Bangladesh.
Latest news
-
Nov 2026: Our paper “Beyond Visible Evidence: Open-World Amodal Counting Under Spatial Occlusion” (CountOCC) has been accepted at the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027)! The work introduces an open-world amodal counting framework that reconstructs occluded object features using spatial context and multimodal guidance. Paper
- Sep 2026: Two new preprints on offline goal-conditioned RL: GTRL, and a study asking whether better goal representations actually help. Read more
- May 2026: PlantPhenoBot, a robotic plant-phenotyping project with Gazipur Agricultural University, begins under a UGC–DU grant.
- Apr 2026: Thinking Like a Botanist is published in Findings of ACL 2026.
- Apr 2026: VLM-Nav, mapless UAV navigation driven by vision-language models, is published in PLOS One.
- Jan 2026: Two papers at IEEE ICECTE 2026: IG-CNN for knee osteoporosis diagnosis and FA-EEGNet for stroke-rehabilitation EEG.
- Oct 2025: A MOST-funded project on explainable AI for histological cancer detection begins with Bangladesh Medical University.
Highlights
Our Research
We study how machines can perceive and act intelligently, from foundations to deployment. Our areas are robot learning, vision-language reasoning, autonomous UAVs, plant phenomics, medical imaging, and intelligent sensing. Each area pairs new methods with a concrete problem worth solving: counting wheat spikes in a field, diagnosing a diseased leaf, explaining a cancer prediction to a pathologist.
Publications
Our recent work spans Findings of ACL, PLOS One, Plant Methods, Smart Agricultural Technology, Journal of Building Engineering, and IEEE Access, alongside new preprints in reinforcement learning and vision-language navigation. Browse every paper since 2022, grouped by year and fully searchable.
Our Projects
Our funded projects are supported by MOST, UGC, and ICT, with partners including Gazipur Agricultural University and Bangladesh Medical University. Running now: PlantPhenoBot, a field-robotics project for plant phenotyping, and explainable AI for cancer pathology. Past projects range from a teleoperated hospital-assistance robot to plant-disease forecasting.
Our Team
AVIS Lab brings together undergraduate and graduate researchers who work closely with Dr. Hasan from first idea to published paper. Students lead their own projects, co-author at international venues, and have been recognized with the Faculty of Engineering and Technology Dean’s Award. Dr. Hasan received one in 2024.