MekaNet: WSI-based Tiny Object Detection
A pathology WSI pipeline for tiny-object detection and clinical-data evaluation under real imaging constraints.
A whole-slide-image pipeline for small object detection in pathology. It belongs to the applied side of my work: turning difficult domain data into usable AI systems.
problem
Whole-slide pathology images are huge, sparse, and clinically constrained, so toy CV pipelines do not translate directly.
key idea
Build a WSI-oriented detection and evaluation pipeline that respects tiny-object scale, hospital data splits, and clinical reporting needs.
my role
Technical contributor; data, evaluation, and manuscript-support role.
methods
- • Whole-slide image processing
- • Tiny-object detection
- • External validation
- • Clinical metric reporting
evidence / results
- • Under review
- • Public release and manuscript-support artifacts exist in the project portfolio
why this belongs in the portfolio
- • Shows the applied-system side of domain-substrate work
- • Adds clinical-data rigor to the portfolio
authors
Jae-Hyun Baek et al.
venue / status
Medical image analysis manuscript
Under review; public-facing details intentionally kept concise.
tags