MediCaption: Integrating YOLO-Driven Computer Vision and NLP for Advanced Pharmaceutical Package Recognition and Annotation
Bibliographic record
Abstract
To ensure patient safety and reduce the incidence of prescription errors, the healthcare industry places a high priority on the availability and accuracy of pharmaceutical information. MediCaption offers a unique solution to this issue with its integrated system, which combines the capabilities of computer vision driven with a state-of-the-art object detection model YOLO-v8 by Ultralytics [1], robust natural language processing (NLP), text-to-speech (TTS), and optical character recognition (OCR). This Project utilizes advanced AI and image processing to quickly and accurately annotate pharmaceutical packaging with key information like drug names, uses, and side effects, significantly reducing medication management errors and enhancing information precision and usability. Using a dataset of 372 pharmaceutical packages from Kaggle (Shah, 2021) [2], we annotated it with Roboflow and trained it using the YOLO-v8 model, achieving precise medicine name detection through bounding box accuracy. This enabled effective text extraction via OCR, following NLP preprocessing by matching against a medicinal database, allowed for the generation of informative captions. To improve user accessibility, these captions were subsequently translated into audio using Text-to-Speech (TTS) technology. This system is designed with computational efficiency and user accessibility in mind, making it beneficial for a wide array of users, including those with visual impairments.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.008 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".