A Novel Data Recovery Technique Using Dactylogram for Identifying Accident Victims

It is challenging to trace the specifics of such fatalities from unforeseen mishaps and occurrences using the technique currently in use. We observe that both the techniques perform well on sample data of about 80 fingerprints. Machine Learning models have been created using Catboost, CNN and Random Forest algorithms and they have been evaluated for metrics like accuracy, precision, recall, F1-score and support and predictions have been made and the results are saved. The accuracy of methods is recorded to be about 96%. It can be observed that the first process is suitable for one-to-one match where as the second process is suitable for batch processing.