Sign in to view source links and access this dataset
Description
An annotated synthetic dataset of 500 Piping and Instrumentation Diagrams (P&IDs) incorporating different types of noise and complex symbols. The dataset, Digitize-PID, was created by Paliwal, S., Jain, A., Sharma, M., & Vig, L. and published in 2021. It contains only symbols, formatted for object detection tasks.
Use Cases
Training object detection models to identify symbols in P&IDs based on the annotated symbols.
Benchmarking model robustness against noise based on the described incorporation of different noise types.
Developing automated tools for converting paper-based engineering schematics to digital formats based on the dataset's focus on digitization.
Researching synthetic data generation for complex, domain-specific visual recognition tasks based on the dataset's synthetic nature.
Strengths
Contains 500 distinct Piping and Instrumentation Diagram images.
Includes annotations specifically for symbol detection tasks.
Incorporates varied noise and complex symbols, which may aid in training robust models.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
The dataset is synthetic, which may not fully capture the distribution of real-world, hand-drawn P&IDs.
Provenance
Source
Original data repository referenced on Hugging Face; primary source is the associated 2021 arXiv paper.
Collection Method
Synthetically generated and annotated, as described.
Freshness
Last updated 2025-08-05 18:17:05; freshness should be verified.
License is unknown; terms of use must be verified before application.