README.md

metadata

language: en
license: mit
tags:
  - semantic-segmentation
  - pytorch
  - transformers
  - segformer
  - agricultural-cv
model-index:
  - name: segformer-b3_run_2026-06-29_00-06_sample_50
    results:
      - task:
          type: semantic-segmentation
          name: Corm Segmentation
        metrics:
          - type: mean_iou
            value: 0.8843405635722065
            name: Best Val Mean IoU
          - type: damage_iou
            value: 0.7566256108147862
            name: Best Val Damage-Class IoU
          - type: root_iou
            value: 0.8987028323778692
            name: Best Val Root-Class IoU
          - type: mean_accuracy
            value: 0.937641067753777
            name: Best Val Mean Accuracy
          - type: damage_accuracy
            value: 0.8839718280719298
            name: Best Val Damage-Class Accuracy
          - type: root_accuracy
            value: 0.9737251750602065
            name: Best Val Root-Class Accuracy

Segformer Corm & Damage Segmentation Model (Ablation: 70.0%)

This repository contains logs for experiment segformer-b3_run_2026-06-29_00-06_sample_50.

Model Summary

• Architecture Type: Segformer (nvidia/segformer-b3-finetuned-ade-512-512)
• Train Sample Fraction: 70.0% - 533/762 images
• Val Sample Fraction: 60.0% - 186/310 images

Metric Curves