Prescription Pill Images for Computer Vision Research

The Computational Photography Project for Pill Identification (C3PI) created the RxIMAGE database of high-quality digital images of prescription pills for computer vision research. Images were captured under controlled laboratory lighting conditions using specialized macro-photography techniques. The project was sunset in 2018, and no new images or identifiers will be added.

Use Cases

• Train image segmentation models on the high-resolution JPEG images to isolate pill features.
• Develop text- and image-based search and retrieval systems using the associated metadata for pill identification.
• Benchmark computer vision algorithms for object recognition on the standardized front and back face pill photographs.

Strengths

• Images were captured under controlled laboratory lighting conditions for consistency.
• Specialized digital macro-photography techniques were used to produce high-resolution images.
• Image segmentation algorithms were applied to create the final JPEG images in the database.

Limitations

• The project was sunset in 2018; data is not updated and may be temporally stale.
• The dataset is for research and development purposes only, not for clinical identification.
• Specific quantitative details like total image count or metadata schema are not provided in the input.

Provenance

Source

U.S. Department of Health & Human Services, National Library of Medicine (NLM) archive.

Collection Method

Photographs taken under laboratory conditions with a camera directly above the front and back faces of pills, using specialized macro-photography.

Time Range

Project active until 2018.

Freshness

The project was sunset in 2018. The platform record was last updated in 2025, but the underlying image collection is static.