Article
AI-Enabled Disaster Prediction and Autonomous Search Operations Platform
During disaster search and rescue operations often hinder because of collapse building debris. Rescue team often suffer to reach victim on time which may cause death to victim. To easy search operations we are combining Robotic Hardware along with Machine Learning algorithms. Robots can easily have reached dangerous locations and then apply Machine Learning algorithm to detect presence of human or their body parts. During disaster ML algorithms must work accurately for human detection so we are employing CNN deep learning algorithm which is best known for optimize features extraction and this extracted feature will get trained with various ML algorithms such as Random Forest, SVM and Decision Tree. Each algorithm performance is evaluated in terms of Accuracy, Precision, Recall, Confusion Matrix and FSCORE. All algorithms able to get more than 90% accuracy on CNN optimized features. For accurate prediction Dataset will play an important role but on internet no such dataset exists so we downloaded some disaster images from internet which is created using Virtual Disaster. Dataset size was so small so we applied augmentation algorithm to increase dataset images count.
Full Text Attachment
