AI-Powered Ballistics: Highlights from Our BALİSTİKA Presentation at IAFS 2026 Sofia!
Finding a "needle in a haystack" is a daily challenge for forensic ballistics teams. Traditional Automated Ballistic Identification Systems (ABIS) excel at 1-to-1 matching, but they often ignore firearm brand signatures—missing a massive opportunity to instantly narrow down search spaces.
At IAFS 2026 in Sofia, Bulgaria, Dr. Ömer Çayırpınar presented key findings from our team's groundbreaking research—authored by one of our lead researchers, Dr. Edanur Meral, and recently published in Forensic Science International under the title "Firearm brand classification using deep learning on cartridge case images".
Solving the "Converted Firearm" & Blank-Pistol Dilemma
In criminal investigations, forensic labs face a growing obstacle: blank-firing pistols illicitly converted to fire live ammunition. These converted handguns and modified weapons create serious forensic challenges:
- ⚠️ Deceptive Toolmarks: Modified or re-engineered barrels and firing mechanisms alter standard toolmark signatures, often confusing traditional matching algorithms and human examiners alike.
- ⚠️ High Volume in Crime: Converted blank pistols make up a massive percentage of illicit firearms recovered from crime scenes.
- 🎯 AI-Driven Resolution: Our dataset explicitly incorporated these converted firearms and blank-pistol variants. By focusing on core primer curvature via 3D Shape Indexing, BALİSTİKA's AI successfully identifies the underlying manufacturer signatures—even when the weapon has been illegally modified!
Proven at IAFS 2026: 91.6% Brand Identification Accuracy
By extracting 3D surface height maps directly from BALİSTİKA, converting them into normalized Shape Index representations, and applying state-of-the-art deep learning (ResNet-50), our team automated brand detection across 350,000+ BALİSTİKA cartridge case scans:
- 91.6% Classification Accuracy: Reliably identifies firearm brand signatures from primer and firing pin impressions captured by BALİSTİKA—including modified and converted models.
- +5% ABIS Search Boost: Integrated into a BALİSTİKA database of nearly 1 million cases, significantly elevating top-candidate matching accuracy.
- Explainable AI: Uses Occlusion Sensitivity heatmaps to prove the AI focuses on forensically meaningful toolmarks—not background noise.
Why BALİSTİKA + AI Gives Your Lab a Competitive Edge
- Accelerated Casework: Automatically filter database queries by brand confidence to cut examiner review times by hours.
- Master Complex Crime Scenes: Confidently classify difficult evidence originating from converted blank pistols and non-standard firearms.
- Database Integrity: AI-driven audits within BALİSTİKA spot existing database mislabelings and legacy entry errors.
- Operational Readiness: Seamlessly supercharges modern 3D ballistics workflows without disrupting established examiner protocols.
Ready to Upgrade Your Forensic Capabilities? 🤝
Missed Dr. Ömer Çayırpınar's session in Sofia? You can read our full paper, "Firearm brand classification using deep learning on cartridge case images," on ScienceDirect or contact our team today to schedule a live demo of BALİSTİKA!
