Skip to content

Latest commit

 

History

History
91 lines (62 loc) · 4.79 KB

README.md

File metadata and controls

91 lines (62 loc) · 4.79 KB
Raw

Post-Disaster Rapid Response Retrofit: PD3R

Natural disasters have been a determinant factor for the development of civilizations. From landscaping the environment to vanishing entire cities, the forces of nature play an important role in our lives. The impact of these events has been significant, especially when they strike on developing countries. For instance, between 1971 and 1995 Natural Disasters took more than 3 million lives, and affected more than 136 million around the world (HRC,2004). Historical evidences show how vulnerable we are to the power of earthquakes, typhoons, hurricanes, and more. However, our mindset towards Natural Disasters has changed through time as modern society has focused more on the scientific facts rather than supernatural explanations. This change of approach has led to the development of tools that can predict, quantify, and reduce risk. As governments and local authorities implement these new tools, life and economic loss after disasters reduce drastically.

Following Build Change's main premise to Build Disaster Resistant Buildings and Change Construction Practices Permanently, PD3R Team's main objective is to improve the safety conditions of buildings and reduce human and economic loss after the occurrence of a natural disaster. Now, this is a broad and ambitious objective, so our main specific objectives can be summarized in the following points:

  • Develop a methodology incorporating AI capabilities to enable rapid post-disaster retrofitting once the disaster strikes.
  • Train the Image Recognition AI to determine if a house is eligible for retrofitting, by feeding it with a large volume of digital images generated through a BIM software, along with some real images.
  • Design the methodology in a way that allows its application globally, adapting the construction types, materials and structural calculations to the context of a particular country or geographic area.
  • Reduce response time to a few days post-disaster, to minimize waiting time to retrofit houses of affected people.

Related GitHub repos

Getting Started

Prerequisites

  • Clone the project from github repo
  • Create virtual environment, if you want using virtualenv command
  • Install dependent libararies
pip install -r requirements.txt

Convert

  • Resize and convert images to the OpenCv's Canny edge filter output
  • real_digital_train folder contains data for training model and real_digital_valid contains data for validating model
  • Final outputs are stored in numpy_edge folder
python convert_data.py

Train

  • Train and make model using train.py file using the output from above step which are stored in numpy_edge folder
  • log directory logs_digital is used for tensorboard to visualize the progress
  • Final model is stored in models folder
python train.py

Test your model

  • Run the below python script to classify your test images based on our pre-trained model and plot confusion matrix .
  • Load model from models folder and test data from numpy_edge folder
python evaluate_confusion_matrix.py

Deployment

"Android Application" folder contains the android code which takes the picture and sends to the AI server and show the final output. To change the api for your own server go to MainActivity.java file and change "BASE_URL" value. ie : public static String BASE_URL = "Your API Code HERE";

Built With

  • Python - The Programming Language used
  • Android - Mobile Application
  • Dynamo - for generating the 3d model

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. Please make sure to update tests as appropriate.

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

Authors

  • Build Change -

See also the list of contributors who participated in this project.

License

This project is licensed under the Apache 2 License - see the LICENSE file for details

Acknowledgments