Pillarox AI - Medicine Identification for Visually Impaired

Pillarox AI is a program made to assist visually impaired individuals with identifying their medication. Trained using Google's Teachable Machine and coded using AI, this project hopes to reduce the health hazards and risks for a visually impaired person.
Rishiraj Panchagnula
STEM Innovation Academy Jr. High SW
Grade 8

Presentation

No video provided

Problem

  • Visually impaired individuals often struggle with taking medication or identifying different medicines. This is because labels and packaging are very reliant on visual information. Misidentifying medication can lead to errors and mistakes, posing a serious health risk and hazard.​

Method

  • Complete background research along with bibliography
  • Start training the model
  • Export model and use AI to code an HTML app
  • Troubleshoot the code for any errors and mistakes
  • Finish building app and gather data
  • Use data to build a graph
  • Create slide deck for trifold
  • Receive trifold and print slide deck
  • Assemble trifold
  • Practice script and revise presenting
  • STEMIA Science Fair on Feb. 20

Analysis

The model reached an average accuracy of 93.3% out of 60 tries. This equates to 56 correct predictions from the AI. One factor that may have affected the accuracy was the Senna S tablets scoring an 80%. This is likely due to the fact that Polysporin and Senna S have a similar color palette.

Conclusion

My project gives visually impaired individuals a safer way of identifying their medication, providing a less risky method than traditional stamps or Braille labels. I aim to implement my project in the real world by improving and working on it more, along with regularly sharing results.

Citations

“What Is Image Classification?” GeeksforGeeks, www.geeksforgeeks.org/computer-vision/what-is-image-classification/#what-is-image-classification. Accessed 15 Nov. 2025. “Accessible Prescription Labels.” American Foundation for the Blind, www.afb.org/blindness-and-low-vision/using-technology/prescription-health-and-fitness/accessible-prescription. Accessed 15 Nov. 2025. Medh, Dipankar. “Real-Time Object Detection with YOLO and Webcam: Enhancing Your Computer Vision Skills.” Medium, dipankarmedh1.medium.com/real-time-object-detection-with-yolo-and-webcam-enhancing-your-computer-vision-skills-861b97c78993. Accessed 15 Nov. 2025. “Confidence.” Ultralytics, www.ultralytics.com/glossary/confidence. Accessed 15 Nov. 2025. Nair, S. S., et al. “Medication Errors in the Visually Impaired Population.” PubMed, pubmed.ncbi.nlm.nih.gov/24513423. Accessed 15 Nov. 2025. “What Is Artificial Intelligence (AI)?” IBM, www.ibm.com/think/topics/artificial-intelligence. Accessed 15 Nov. 2025. “Machine Learning.” GeeksforGeeks, www.geeksforgeeks.org/machine-learning/. Accessed 15 Nov. 2025. Teachable Machine. Google, teachablemachine.withgoogle.com/. Accessed 15 Nov. 2025. TensorFlow.js. Google, www.tensorflow.org/js. Accessed 15 Nov. 2025. “How TO - Speech Synthesis.” W3Schools, www.w3schools.com/howto/howto_js_speech_synthesis.asp. Accessed 15 Nov. 2025. “AI for Accessibility.” Microsoft, www.microsoft.com/en-us/ai/ai-for-accessibility. Accessed 15 Nov. 2025.

Acknowledgement

I would like to acknowledge my mom for helping me gather supplies for my trifold and motivating me not to quit. I would also like to thank my sister for supporting me and giving me feedback on my project.