Litcius/Paper detail

Advantages and Limitations of Fully on-Chip CNN FPGA-Based Hardware Accelerator

Gianmarco Dinelli, Gabriele Meoni, Emilio Rapuano, Luca Fanucci

202035 citationsDOI

Abstract

Convolution Neural Networks are a class of deep neural networks commonly used in audio and video elaborations. Their implementation on the edge represents a complex task due to the limited computational power and low power consumption requirement that characterize these applications. In this paper, a fully on-chip Convolutional Neural Network Field Programmable Gate Array-based hardware accelerator is presented. This approach allows to reduce power consumption due to off-chip memory accesses and aims to reduce design time. Advantages and limitations of the proposed architecture are discussed and a trade-off analysis is provided to give intuitions about the feasibility of this method.

Topics & Concepts

Computer scienceField-programmable gate arrayConvolutional neural networkEmbedded systemChipConvolution (computer science)Computer hardwarePower consumptionHardware accelerationSystem on a chipArtificial neural networkTask (project management)Computer architecturePower (physics)Artificial intelligenceEngineeringSystems engineeringPhysicsTelecommunicationsQuantum mechanicsAdvanced Neural Network ApplicationsHuman Pose and Action RecognitionAdvanced Vision and Imaging