The FASoC Datasheet Scrubber is a utility that scrubs through large sets of PDF datasheets/documents in order to extract key circuit information. The information gathered is used to build a database of commercial off-the-shelf (COTS) IP that can be used to build larger SoC in the FASoC design. More information here.
To get more details about the datasheet scrubber, please refer to our IEEE TCAD paper.
If you find this tool useful in your research, we kindly request to cite our paper below:
-
M. Fayazi et al., "FASCINET: A Fully Automated Single-Board Computer Generator Using Neural Networks," in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.
-
Z. Colter et al., "Tablext: A combined neural network and heuristic based table extractor," in Array,Vol. 15, 2022, pp. 100220.
-
Ensure your machine has the correct Python version and all of the Python modules required to run through the datasheet scrubber.
- Requirements: Anaconda, and Tensorflow. Python versions below 3.6 are not supported.
-
Ensure you have ssh keys setup for github. Instructions for generating and adding ssh keys can be found here.
-
Clone the Datasheet Scrubber repository
git clone git@github.com:idea-fasoc/datasheet-scrubber.git
The FASoC database contains more than 700,000 records of Integrated Circuits (ICs) components collected from Digikey.
In order to access a sample of this collection, visit our web application or proceed here.
To have access to the entire collection of components, please visit here.
Datasheet scrubber includes three steps of category recognition, table extracton and text extraction.
Examples of how to use the category recognition, table extractor, and web application database are provided.