Website deploying models for
- Detecting damage in images of destruction caused by disasters like Earthquakes/Tsunamis.
- Classifying the severity of damage in 5 classes. User can submit images and have severity analysis as well as detection results.
Computer Vision model YOLO (You Only Look Once) v7 is trained and deployed For inference, weights file 'best.pth' is used and the github repository https://github.com/WongKinYiu/yolov7 must be cloned
The manage.py must be execited through terminal, and the link provided for the local host is to used with the extention '../homepage'
The smaller Dataset :- https://universe.roboflow.com/ai-imagery/isbda_classification
The larger Dataset:- https://universe.roboflow.com/kios-alpsc/rescuenet-yolo8/browse?queryText=&pageSize=50&startingIndex=0&browseQuery=true
The given Dataset https://eod-grss-ieee.com/dataset-detail/OWlUN0k3T0tnNVo2Rk5mNjFxTkhwZz09
Anany Garg, Dyuti Ballav Paul, Prachi Vakshi, Pranav Gupta, Samyak Jha