Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/nickklos10/concrete-crack-detector-cv
A web application for detecting cracks in concrete surfaces using a deep learning model integrated with Flask.
https://github.com/nickklos10/concrete-crack-detector-cv
axios flask html-css-javascript jinja2 pillow pytorch torchvision
Last synced: about 2 months ago
JSON representation
A web application for detecting cracks in concrete surfaces using a deep learning model integrated with Flask.
- Host: GitHub
- URL: https://github.com/nickklos10/concrete-crack-detector-cv
- Owner: nickklos10
- Created: 2024-11-04T21:51:58.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2024-11-11T22:37:58.000Z (about 2 months ago)
- Last Synced: 2024-11-11T23:28:49.558Z (about 2 months ago)
- Topics: axios, flask, html-css-javascript, jinja2, pillow, pytorch, torchvision
- Language: Jupyter Notebook
- Homepage:
- Size: 39.7 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Concrete Crack Detection Web App
A Flask-based web application to detect cracks in concrete using a pre-trained ResNet-18 model. The application allows users to upload images of concrete surfaces and utilizes a deep learning model to determine whether a crack is present with a high level of confidence.
## Table of Contents
- [Project Overview](#project-overview)
- [Technologies Used](#technologies-used)
- [Installation](#installation)
- [Usage](#usage)
- [Model Training Details](#model-training-details)
---
## Project Overview
This project provides a web-based interface for detecting cracks in concrete surfaces, leveraging a trained deep learning model built using ResNet-18. Users can upload images for real-time prediction, with results indicating the presence or absence of a crack and the confidence level of the prediction.
## Technologies Used
- **Flask** - for creating the web application and handling HTTP requests
- **PyTorch** - for model development and training, utilizing the ResNet-18 architecture
- **Torchvision** - for image transformations and pre-trained model utilities
- **PIL (Pillow)** - for handling image uploads and conversions
- **HTML, CSS, JavaScript** - for building the user interface and handling client-side interactions
- **Axios** - for making HTTP requests from the frontend to the backend
- **Jinja2** - for template rendering in Flask
## Installation
1. **Clone the Repository:**
```
git clone https://github.com/nickklos10/Concrete-Crack-Detector-CV.git
cd Concrete-Crack-Detector-CV
```
2. **Install Dependencies:** Ensure you are using a Python virtual environment, then install the required packages:
```
pip install -r requirements.txt
```
3. **Add the Pre-trained Model:**
- Place the resnet18_trained.pth model file in the project directory. You may need to train the model yourself if it’s not provided due to size constraints (see Model Training Details).
4. **Run the Application:**
```
python app.py
```
The app should be accessible at http://localhost:8000.
## Usage
1. Open the web app in a browser by navigating to http://localhost:8000.
2. Drag and drop an image or click to upload a concrete surface image.
3. Wait for the model to process the image, and view the results with the crack detection status and confidence level.
## Model Training Details
The deep learning model is based on ResNet-18, a convolutional neural network architecture well-suited for image classification tasks.
## Training Script
The model was trained using a custom training script in PyTorch. Here’s an overview of the training process:
1. **Dataset Preparation:**
A labeled dataset of concrete images with and without cracks was used. Each image was resized to 227x227 pixels for compatibility with the ResNet-18 model.
2. **Data Transformations:**
Applied standard image transformations, including resizing, normalization with mean [0.485, 0.456, 0.406], and standard deviation [0.229, 0.224, 0.225].
3. **Model Architecture:**
ResNet-18 was used as the base architecture. The final fully connected layer was modified to output two classes: "Crack" and "No Crack."
4. **Training Parameters:**
Optimizer: `Adam`
Learning Rate: `0.001`
Loss Function: `CrossEntropyLoss`
Number of Epochs: `20`
Batch Size: `32`
5. **Saving the Model:**
- After training, the model was saved using torch.save:
```
torch.save(model.state_dict(), 'resnet18_trained.pth')
```
## Frontend Look of the Web-App
