Örnek, Ahmet HaydarÇelik, MustafaAlper, Ozan Can2022-05-232022-05-2320219781665442312https://doi.org/10.1109/ICECET52533.2021.9698252https://hdl.handle.net/20.500.13091/23792021 International Conference on Electrical, Computer, and Energy Technologies, ICECET 2021 -- 9 December 2021 through 10 December 2021 -- -- 177176In order to realize real-time computer vision projects we need to avoid time consuming operations such as more inference for deep learning. Our current application uses face images to decide whether there is a mask on the face so as to prevent unhealthy situations in view of epidemic. Since frames are sequentially coming it is necessary to eliminate similar frames to avoid more inference. We show how to measure a similarity between two frames by comparing traditional and deep learning based methods in this study. This study shows that deep learning based method is more efficient than traditional methods when comparing images. © 2021 IEEE.eninfo:eu-repo/semantics/closedAccessdeep learningimage similarityreal-world applicationsDeep learning'currentDeep learningFace imagesImage similarityLearning-based methodsReal-time computer visionReal-worldReal-world applicationImage analysisConference Object10.1109/ICECET52533.2021.96982522-s2.0-85127068598