AIbase
EN
Home
Ecc Segformerv1
E

Ecc Segformerv1

Developed by rishitunu
Image segmentation model fine-tuned on nvidia/mit-b5, specializing in crack detection tasks
Image Segmentation
Transformers
Open Source License:Other#Concrete crack detection#High-precision segmentation#Building structure assessment
Downloads 22
Release Time : 8/8/2023

Model Overview

This model is an image segmentation model fine-tuned on the rishitunu/ecc_crackdetector dataset based on the nvidia/mit-b5 architecture, primarily used for crack detection tasks. It demonstrates high Intersection over Union (0.9171) and accuracy (0.8041) on the evaluation set.

Model Features

High-precision crack detection
Achieves 0.9171 Intersection over Union and 0.8041 accuracy in crack detection tasks
Based on SegFormer architecture
Uses nvidia/mit-b5 as the base model, combining the advantages of Transformer in vision tasks
Lightweight training
Trained with a small batch size (2), suitable for resource-constrained environments

Model Capabilities

Image segmentation
Crack detection
Visual analysis

Use Cases

Infrastructure inspection
Concrete crack detection
Used to detect cracks in concrete structures of buildings or bridges
Achieves 91.71% Intersection over Union on the test set
Industrial inspection
Material surface defect detection
Detects cracks and defects on industrial material surfaces
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase