A new study published in the Journal of Forensic Science suggests artificial intelligence may soon streamline forensic cannabis identification, replacing slower traditional methods.
For the study, researchers from the Israel Institute of Technology developed a deep learning system capable of detecting cannabis by analyzing microscopic images of plant trichomes—specifically, non-glandular trichome hairs, a distinguishing feature of marijuana. The study notes that conventional approaches typically rely on chemical tests like Duquenois-Levine and Fast Blue BB, along with expert microscopic review, making the process labor-intensive and slow.
Using a dataset containing thousands of annotated microscope images, including samples of genuine cannabis and non-cannabis materials laced with synthetic cannabinoids, the AI model was trained to recognize microscopic features unique to cannabis. The dataset’s accuracy was verified using chemical assays and expert analysis.
The AI system achieved a classification accuracy above 97%, offering a fast and precise alternative to expert testing. The authors suggest this approach could save time and resources in forensic labs, while bolstering efforts to combat counterfeit marijuana products and illicit trafficking.
More articles from The Marijuana Herald:
