ICVRL International Computer Vision Research Laboratory

Image Segmentation

Pixel-level understanding of images through semantic, instance, and panoptic segmentation techniques.

1 Researchers 1 Projects 2 Publications

Research Focus & Scope

Image Segmentation research at ICVRL delivers dense, pixel-level visual understanding. We develop cutting-edge semantic, instance, and panoptic segmentation models that delineate complex boundaries in high-resolution visual feeds with millimeter precision.

How We Conduct Research in Image Segmentation

Leveraging transformer-based mask decoders, feature pyramid networks (FPN), boundary-aware loss functions, and self-supervised pre-training, our algorithms produce clean, sharp masks even across fine structures, occluded objects, and noisy backgrounds.

Real-World Applications

Precision agriculture crop and weed delineation
Autonomous driving drivable area and obstacle segmentation
Surgical navigation and organ boundary mapping
Satellite land cover and water body mapping
Digital photo foreground/background separation

Projects in this Domain

All Projects

MedSeg-AI: Intelligent Medical Image Segmentation Platform

An AI-powered platform for automated segmentation of medical images including CT, MRI, and X-ray scans with cl...

active Code

Recent Publications

All Publications
conference

Efficient Panoptic Segmentation for Autonomous Driving Scenes

Ahmad Khan, Hassan Ahmed, Ali Raza, Shams ur Rehman

International Conference on Computer Vision (ICCV), 2025

journal

Deep Attention-Guided Network for Medical Image Segmentation

Shams ur Rehman, Ahmad Khan, Muhammad Hamza

IEEE Transactions on Medical Imaging, 2025

Tech Stack & Tools

Mask R-CNN Segment Anything (SAM) DeepLabV3+ UNet++ MMSegmentation PyTorch OpenCV

Join Our Research

Interested in conducting MS/PhD research or undergraduate projects in Image Segmentation?

Apply to Join ICVRL

Explore Other Research Areas

All Areas