VisionIdentify GPT: How well can it detect biological images?
Related Experiment Video
Updated: Nov 4, 2026
In Vivo Imaging Systems (IVIS) Detection of a Neuro-Invasive Encephalitic Virus
Published on: December 2, 2012
1Associate teaching professor, Mike Petryk School of Dentistry, Faculty of Medicine & Dentistry, College of Health Sciences, University of Alberta, Edmonton, AB, Canada.
Introduction:
The application of advanced neural networks has equipped the Generative Pretrained Transformer (GPT-4) with the ability to "see," "hear," and "speak." This new GPT, also known as "VisionIdentify GPT," can now detect patterns and extract meaningful information from visual data.
Project Description:
This study explored how well VisionIdentify GPT can recognize biological images and help develop image-based teaching materials for dental and medical education. Hand-drawn biological illustrations and histological sections were used as inputs for VisionIdentify GPT.
Discussion:
The new GPT showed high accuracy in identifying biological illustrations. However, it was inaccurate in detecting histological sections and was unable to annotate images accurately. VisionIdentify GPT has high potential but has yet to be developed fully.
Conclusions:
Specialized training with discipline-specific data can enhance the mastery of this new vision-powered GPT, potentially making it a helpful tool for specialized tasks.
