This article presents a workflow for unsupervised clustering a large collection of forensic images.
Large collections of images, if curated, drastically contribute to the quality of research in many domains. Unsupervised clustering is an intuitive, yet effective step towards curating such datasets. The workflow in the current project utilizes classic clustering on deep feature representation of the images in addition to domain-related data to group them together. The manual evaluation shows a purity of 89% for the resulted clusters. (Published abstract provided)
Similar Publications
- Further Development of Raman Spectroscopy for Body Fluid Investigation: Forensic Identification, Limit of Detection, and Donor Characterization
- Raman Spectroscopy and Chemometrics for Forensic Bloodstain Analysis: Species Differentiation, Donor Age Estimation, and Dating of Bloodstains
- Targeted-Ion Mass Spectrometry for the Identification of Forensically Relevant Biological Fluids and Samples from Sexual Assault Evidence