Automated shape annotation for illicit tablet preparations: a contour angle based classification from digital images.

Area of Science:

• Forensic Science
• Computer Vision
• Data Analysis

Background:

• Managing large datasets of physical evidence, such as illicit tablets, is a challenge for forensic intelligence.
• Automated methods are needed to efficiently analyze and categorize seized items.

Purpose of the Study:

• To develop an automated shape classification method for illicit tablets.
• To facilitate forensic intelligence by enabling categorical shape annotation of seized tablets.

Main Methods:

• The approach uses digital images of seized tablets and is invariant to scale, rotation, and translation.
• It involves two processing levels: coarse level comparing contour curvature space and fine level using a classification tree.
• The method performs categorical shape annotation for illicit tablets.

Main Results:

• The automated method was tested on a collection of 169 tablets with diverse shapes.
• An accuracy of 97.6% was achieved when defining 19 distinct shape categories.
• The system effectively categorizes tablet shapes for forensic analysis.

Conclusions:

• The developed automated shape classification method is highly accurate and efficient.
• This tool can significantly enhance forensic intelligence efforts in managing physical feature data of illicit tablets.
• The approach offers a robust solution for objective and consistent tablet shape categorization.