Global interlaboratory study: Assessing consistency and accuracy for determining source of face, hands, and clothing...

Images containing faces, body parts, vehicles, weapons, clothing, luggage, furniture, landscapes, and buildings are routinely sent to forensic laboratories for comparison. Trained forensic practitioners offer an opinion regarding a common source or a different source. Results from comparisons are typically reported using opinion categories, and the number used can vary (typically 3-7 categories). To harmonize opinion categories used across digital multimedia disciplines, the Organization of Scientific Area Committees (OSAC) for Forensic Science developed OSAC Proposed Standard 2022-S-0001 (Standard Guide for Image Comparison Opinions), which details five opinion categories for determining image source. To ensure this standard meets the OSAC program goal of developing technically sound standards, an interlaboratory study focused on 2022-S-0001 was conducted. Combining machine learning and input from subject matter experts, 20 pairs of face, hands, and clothing images were selected for use that represented varying sources and difficulty. Qualtrics was used to disseminate a survey and collect responses from forensic practitioners (n = 38) and laypersons (n = 44). For all three image subject categories, statistically higher agreement (selection of the same opinion source category for the same pair of images) was observed between practitioners over laypersons. Practitioners were statistically more accurate at determining the true source over laypersons for face (p = 0.005) and hand (p = 0.043) image pairs. This study highlighted that when individuals are provided with a clear and detailed framework of source opinion categories, moderate agreement and high accuracy (especially for practitioners) can be achieved for face, hand, and clothing image comparisons.