New solutions for face photo retrieval based on sketches

Kukharev, Georgy; Matveev, Yuri; Shchegoleva, Nadezhda

doi:10.1134/s1054661816010144

Cited by 6 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…EvoFIT composites constructed under laboratory conditions have seen accuracy levels improve from a few percent correct [5,6] to around 35 to 45% [23,24], and higher [18]. At first glance, the lower naming levels found here (both Conditions in Experiment 1, and Witness At Home in Experiment 2) may give cause for concern.…”

Section: Final Commentsmentioning

confidence: 71%

“…immediately or within 3-4 hours) after target face encoding are usually named at around 20% correct [4]. However, the introduction of a forensically relevant delay, in the region of 24 to 48 hours, results in worse performance [5,6]. The development of holistic (or recognition-based) systems (e.g., EvoFIT) with enhanced interviewing techniques (e.g.…”

Section: Discussionmentioning

confidence: 99%

“…Under certain circumstances, in particular where composite construction is carried out immediately or within a few hours of viewing a target, "feature" systems result in composites with mean correct naming of up to 20% [4]. However, inclusion of a forensically-relevant delay of upwards of 24-48 hours results in composites with mean naming of only a few percent correct (e.g., [5,6].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

EvoFIT composite face construction via practitioner interviewing and a witness-administered protocol

Martin

Jaeckel

Frowd

et al. 2018

2018 NASA/ESA Conference on Adaptive Hardware and Systems (AHS)

View full text Add to dashboard Cite

Police require reliable facial-composite systems to help identify, arrest and convict criminals. Recent developments, however, have allowed newer versions of the EvoFIT composite system to be made available for policing. The outcome is an online (cloudbased) version and a new system called Witness At Home, both using a simpler interface. Here, we formally compare these two versions to establish potential benefits to policing. Two experiments were conducted. In Experiment 1, participants observed a target identity for 1 minute and returned 4 hours (Witness At Home) or 24 hours (EvoFIT Online) to construct a composite from memory. No significant difference in composite accuracy was found. In Experiment 2, participants constructed a composite, 24-hours after seeing a target identity, using either EvoFIT Online or Witness At Home. A significant increase in accurate identification was found for EvoFIT Online, with some utility for the self-administered procedure, together indicating benefit for these newer systems plus some areas for development.

show abstract

Section: Final Commentsmentioning

confidence: 71%

Section: Discussionmentioning

confidence: 99%