decimer.ai

What It Is

DECIMER.ai is an open web application for scientific publications, focused on automatically identifying, segmenting, and parsing chemical structure diagrams. Users can upload PDF documents or one or more images containing chemical structures. The system detects structure diagrams in the literature and converts them into computable chemical representations using the DECIMER V2 OCSR engine.

Core Capabilities

Its AI capabilities are centered on chemical image understanding: DECIMER Segmentation detects and crops chemical structures from PDFs, while DECIMER Image Classifier and DECIMER Transformer support structure classification and recognition, ultimately generating SMILES. The page also notes that STOUT V2 can be used to parse IUPAC names, and that results can be manually corrected in the Ketcher chemical structure editor before downloading segmented images and mol files. The overall workflow is well suited to literature mining, database construction, and organizing structures in drug discovery.

Pricing and Ease of Use

The main text does not disclose subscription pricing, paid plans, usage limits, or trial rules. It only describes the service as an open platform, so it can be considered publicly accessible, but no further assumptions can be made about commercial licensing. The interaction model is straightforward, supporting drag-and-drop, file browsing, and pasted files, making it suitable for researchers who need to quickly process images or PDFs. However, there is no visible information about an API, SDK, batch jobs, file size limits, or similar details, so its suitability for large-scale automated integration is unclear.

Privacy and Limitations

The privacy description is relatively clear: uploaded data is used only for automated processing and result display, and is usually stored for one hour. If a user reports an issue, the relevant images are used for error analysis and then deleted. The site also uses Google Analytics for traffic statistics. One important caveat is that users must ensure they have data-mining rights granted by the publisher for any papers or images they upload. In terms of recognition quality, the page does not provide accuracy figures or define its applicable boundaries, and the inclusion of Ketcher editing suggests that results may still require manual review.

Best Fit and Access from China

DECIMER.ai is best suited to cheminformatics researchers, scientific literature data-mining users, drug discovery teams, and anyone who needs to convert chemical structure diagrams from papers into SMILES/mol formats. It is not intended for general-purpose OCR or non-chemical use cases. Access from mainland China, network stability, and payment methods are not mentioned in the main text and would need to be tested directly. If access is unstable, alternatives such as ChemDraw, OSRA, MolVec, or local RDKit-based workflows may be worth considering.