Connecting national flags – a deep learning approach

Abstract National flags are the most recognizable symbols of identity a country. Similarities between may be observed due to cultural, historical, or ethical connections nations, because they originated from same group people, unrelated sharing common and colors. Although fact that similar exist is indisputable, this has never been quantified. Quantifying flags’ similarities could provide useful body knowledge for vexillologists historians. To end, work aims develop supporting tool scientific study nations’ history symbolisms, through quantification varying degrees similarity their flags, by considering three initially stated hypotheses using novel feature inclusion (FI) measure. The proposed FI measure objectively quantify overall based on optical multi-scaled features extracted flag images. State-of-the-art deep learning models built other applications tested capability first time problem under transfer learning, towards calculating More specifically, was quantified six models: Yolo (V4 V5), SSD, RetinaNet, Fast R-CNN, FCOS CornerNet. Flags’ images dataset included 195 nations officially recognized United Nations. Experimental results reported maximum up 99%. were subsequently justified with help Vexillology domain, support research findings raise questions further investigation. reveal approach reliable able serve as social sciences extraction quantification.