The IRMA Community
Newsletters
Research IRM
Click a keyword to search titles using our InfoSci-OnDemand powered search:
|
Deep Learning-Based Multimodal Biometric Fusion for Forensic Person Identification in Challenging Environments
Abstract
This chapter summarizes, analyses and discusses the current maturity of deep learning technology regarding the development and use of multimodal biometric fusion for forensic identity verification. The chapter identifies several challenges in relation to the collection and processing of wild data (low quality/resolution images, noise due to environmental factors and/or partial occlusion). The chapter discusses advantages of using deep neural networks (DNNs) to replace traditional single-mode systems for the processing and recognition of wild data sets. Finally, the chapter describes three primary methods of fusion (feature-level, score-level and attention-based) that can be effectively used to combine disparate data sets; stresses the importance of cross-modal transformers for achieving accurate identification despite severe degradation caused by the use of dissimilar data sets; presents and emphasizes several advanced preprocessing methods, including the use of generative adversarial networks (GAN) for super-resolution.
Related Content
|
Kavita Kanwar, Nikhil Kumar Goyal.
© 2026.
30 pages.
|
|
Deepak Gupta, Raghu Nangunuri, Srinivasan Nagaraj, S. Keerthi, Pratish Rawat, C. Umarani, Someshwar Siddi.
© 2026.
30 pages.
|
|
Arun Agrawal.
© 2026.
22 pages.
|
|
Aditya Ojha, Sneha Singh, Jyoti Singh Kirar.
© 2026.
50 pages.
|
|
Prachi Sharma Biswas, Swati Dubey Mishra.
© 2026.
34 pages.
|
|
Tamara Phillips Fudge.
© 2026.
34 pages.
|
|
Bayram Cadıl, Gurkan Tuna.
© 2026.
34 pages.
|
|
|