Conceitos essenciais
IDTrust introduces a deep-learning framework for assessing the quality of identification documents, enhancing dataset applicability and accuracy in distinguishing between original and scanned IDs.
Estatísticas
"MIDV-2020 includes 1000 video clips, 2000 scanned images, and 1000 photos of 1000 unique dummy IDs."
"L3i-ID consists of 17 types of original French IDs, comprising 5 identity cards in the old format, 2 identity cards in the new format, 6 passports, and 4 driving licenses."
"The batch size is set to 8, and the number of epochs is 100."
Citações
"IDTrust eliminates the need for original document patterns and pre-processing, improving dataset applicability."
"GuidedDeepQD consistently outperforms DeepQD on the L3i-ID dataset, achieving higher accuracy and F1 scores."