Project

TomatoMAP Research Lane

Guarded public research surface for plant-disease AI and computer vision, with careful dataset provenance and benchmark-safe wording.

TomatoMAP is a public research lane connected to my PhD direction in plant-disease AI and computer vision.

The public repository is framed as a downstream workflow and benchmark artifact archive. It is not the canonical upstream dataset repository, and public writing should preserve that distinction.

Safe Public Framing

Safe wording:

PhD research on label-efficient tomato instance segmentation using TomatoMAP-Seg, with benchmark-first evaluation of supervised baselines under reduced annotation budgets.

What This Proves

This project is useful as evidence of:

  • benchmark discipline;
  • generated artifact tracking;
  • careful dataset provenance;
  • research workflow organization;
  • public/private boundary awareness.

Public Boundary

The public site can mention the broad research area, the benchmark discipline, and the existence of reproducible workflow artifacts. It should not describe exact unpublished methods, private experiment plans, or claims that are not yet supported by final results.

What To Avoid

Avoid public claims about:

  • field deployment;
  • yield improvement or treatment recommendations;
  • broad SOTA claims before final evidence;
  • future weak-label or pseudo-label experiments as if completed;
  • exact unpublished thesis methods.