Litcius/Paper detail

CoDeNet: Efficient Deployment of Input-Adaptive Object Detection on Embedded FPGAs

Qijing Huang, Dequan Wang, Zhen Dong, Yizhao Gao, Yaohui Cai, Tian Li, Bichen Wu, Kurt Keutzer, John Wawrzynek

202152 citationsDOIOpen Access PDF

Abstract

Deploying deep learning models on embedded systems for computer vision tasks has been challenging due to limited compute resources and strict energy budgets. The majority of existing work focuses on accelerating image classification, while other fundamental vision problems, such as object detection, have not been adequately addressed. Compared with image classification, detection problems are more sensitive to the spatial variance of objects, and therefore, require specialized convolutions to aggregate spatial information. To address this need, recent work introduces dynamic deformable convolution to augment regular convolutions. Regular convolutions process a fixed grid of pixels across all the spatial locations in an image, while dynamic deformable convolution may access arbitrary pixels in the image with the access pattern being input-dependent and varying with spatial location. These properties lead to inefficient memory accesses of inputs with existing hardware.

Topics & Concepts

Computer scienceConvolution (computer science)PixelField-programmable gate arrayArtificial intelligenceComputer visionProcess (computing)Object (grammar)Object detectionGridImage (mathematics)Software deploymentAggregate (composite)Image resolutionVariance (accounting)Computer engineeringPattern recognition (psychology)Computer hardwareMathematicsAccountingComposite materialBusinessMaterials scienceArtificial neural networkGeometryOperating systemAdvanced Neural Network ApplicationsCCD and CMOS Imaging SensorsIndustrial Vision Systems and Defect Detection
CoDeNet: Efficient Deployment of Input-Adaptive Object Detection on Embedded FPGAs | Litcius