Using double attention for text tattoo localisation

Text tattoos contain rich information about an individual for forensic investigation. To extract this information, text tattoo localisation is the first and essential step. Previous studies applied existing object detectors to detect general tattoos, but none of them considered they neglect prior knowledge that are usually inside or nearby larger appear only on human skin. use knowledge, a knowledge-based attention mechanism (PKAM) network named Tattoo Localisation Network based Double Attention (TTLN-DA) proposed. In addition TTLN-DA, two variants TTLN-DA designed study effectiveness different knowledge. For study, NTU V2, largest dataset V1, established. examine importance proposed networks, its compared with state-of-the-art detectors. The experimental results indicate vital localisation; PKAM contributes significantly performance outperforms scene