Disaster Classifier

• Task: Development of customized disaster media service platform technology for disaster information and vulnerable groups (재난정보인지 취약계층을 위한 맞춤형 재난미디어 서비스 플랫폼 기술 개발)
• Task has been completed by VI Lab (Video Inteligence Lab) (~12. 31, 2023) with ETRI (Electronics and Telecommunications Research Institute)
• Automatic Classification of Disaster Images Based on Deep Learning [Paper] published in KCIS (The Korean Institute of Communications and Information Sciences, 한국통신학회)
• Patent application (딥러닝 기반 이미지 데이터 분류 장치 및 방법)
• Record Disaster classifier task in Notion [Link]

Start

Python 3.9+

pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch-nightly -c nvidia
conda install pandas, numpy, tqdm, timm, albumentation

• Prepare dataset (disaster or non-disaster images)
• Train, Evaluation code : train.py, eval.py
• Classifier model (inception_resnet-v2) [Google drive]

Proposed Method

• Augment disaster images

  

• Framework