A library-based tool to translate high level DNN models into hierarchical VHDL descriptions
View/ Open
Identifiers
URI: http://hdl.handle.net/10317/10367Share
Metrics
Statistics
View Usage StatisticsMetadata
Show full item recordResearch Group
Desarrollo de sistemas y circuitos electrónicos y microelectrónicosKnowledge Area
Arquitectura y Tecnología de ComputadorasSponsors
This work has been partially funded by Spanish Ministerio de Ciencia e Innovación (MCI), Agencia Estatal de Investigación (AEI) and European Region Development Fund (ERDF/FEDER) under grant RTI2018-097088-B-C33.Realizado en/con
Universidad Politécnica de CartagenaPublication date
2021-11-25Bibliographic Citation
P. Rubio-Ibáñez, J. Javier Martínez-Álvarez, G. Doménech-Asensi, “A library-based tool to translate high level DNN models into hierarchical VHDL descriptions”, XXXVI Conference on Design of Circuits and Integrated Systems, November 24-26th 2021, Vila do Conde, Portugal.Peer review
SíKeywords
High-Level SynthesisVHDL
Deep Neural Network
Keras
Tensorflow
Abstract
This work presents a tool to convert high level models of deep neural networks into register transfer level designs. In order to make it useful for different target technologies, the output designs are based on hierarchical VHDL descriptions, which are accepted as input files for a wide variety of FPGA, SoC and ASIC digital synthesis tools. The presented tool is aimed to speed up the design and synthesis cycle of such systems and provides the designer with certain capability to balance network latency and hardware resources. It also provides a clock domain crossing to interface the input layer of the synthesized neural networks with sensors running at different clock frequencies. The tool is tested with a neural network which combines convolutional and fully connected layers designed to perform traffic sign recognition tasks and synthesized under different hardware resource usage specifications on a Zynq Ultrascale+ MPSoC development board.
Collections
- Artículos [1169]
The following license files are associated with this item:
Social media