CoolFace
Modelpublic

AIdeveric/llava-pavement-distress-lora

sourceHugging Facemitupdated 6mo agoView on Hugging Face
0likes8downloads
Model Card

🛣️ LLaVA-1.5-7B LoRA Adapter - Pavement Distress Analysis

![License: MIT](https://opensource.org/licenses/MIT) ![Entity F1](https://huggingface.co/AIdeveric/llava-pavement-distress-lora) ![Base Model](https://huggingface.co/llava-hf/llava-1.5-7b-hf) ![Open In Colab](https://colab.research.google.com/github/AIdeveric/llava-pavement-distress/blob/main/LLaVAStructuralHealthMonitoringPavementDistressAnalyzer.ipynb)

AI-powered pavement distress analysis using LLaVA-1.5-7B with LoRA fine-tuning.

This repository contains the LoRA adapter achieving 74.31% Entity F1 on pavement distress detection, severity assessment, and maintenance recommendation.

🚀 Quick Start - Try the Demo!

📓 Download and run the Gradio demo notebook:

  1. 1.Download the notebook: `LLaVA_Structural_Health_Monitoring_Pavement_Distress_Analyzer.ipynb`
  1. 1.Upload to Google Colab:
  2. 2.Go to Google Colab
  3. 3.Click File → Upload notebook
  4. 4.Select the downloaded .ipynb file
  1. 1.Run all cells:
  2. 2.Click Runtime → Run all
  3. 3.Wait 3-5 minutes for model to load
  4. 4.Click the public Gradio link to start analyzing images!
  1. 1.Test with sample images:
  2. 2.Download sample images from `/images` folder
  3. 3.Upload to the Gradio interface
  4. 4.Get instant distress analysis!

✅ No setup required - runs entirely in the cloud!