Similarity-aware CNN for efficient video recognition at the Edge
Similarity-aware CNN for efficient video recognition at the Edge
Convolutional neural networks (CNNs) often extract similar features from successive video frames due to having identical appearances. In contrast, conventional CNNs for video recognition process individual frames with a fixed computational effort. Each video frame is independently processed, resulting in numerous redundant computations and an inefficient use of limited energy resources, particularly for edge computing applications. To alleviate the high energy requirements associated with video frame processing, this paper presented similarity-aware CNNs that recognise similar feature pixels across frames and avoid computations on them. First, with a loss of less than 1% in recognition accuracy, a proposed similarity aware quantization technique increases the average number of unchanged feature pixels across frame pairs by up to 85%. Then, a proposed similarity-aware dataflow improves energy consumption by minimising redundant computations and memory accesses across frame pairs. According to simulation experiments, the proposed dataflow decreases the energy consumed by video frame processing by up to 30%.
Computational modeling, Convolutional neural networks, Deep neural networks, Energy consumption, Memory management, Object Detection, Quantization, Quantization (signal), System-on-chip, Tensors, Video Recognition.
Sabetsarvestani, Mohammadamin
f5c0e55f-6f0c-4f56-9d6d-7de19d6fb136
Hare, Jonathon
65ba2cda-eaaf-4767-a325-cd845504e5a9
Al-Hashimi, Bashir
bfee994d-8c63-4fe7-8ec7-76680eb1b642
Merrett, Geoff
89b3a696-41de-44c3-89aa-b0aa29f54020
20 December 2021
Sabetsarvestani, Mohammadamin
f5c0e55f-6f0c-4f56-9d6d-7de19d6fb136
Hare, Jonathon
65ba2cda-eaaf-4767-a325-cd845504e5a9
Al-Hashimi, Bashir
bfee994d-8c63-4fe7-8ec7-76680eb1b642
Merrett, Geoff
89b3a696-41de-44c3-89aa-b0aa29f54020
Sabetsarvestani, Mohammadamin, Hare, Jonathon, Al-Hashimi, Bashir and Merrett, Geoff
(2021)
Similarity-aware CNN for efficient video recognition at the Edge.
IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 41 (11).
(doi:10.1109/TCAD.2021.3136815).
Abstract
Convolutional neural networks (CNNs) often extract similar features from successive video frames due to having identical appearances. In contrast, conventional CNNs for video recognition process individual frames with a fixed computational effort. Each video frame is independently processed, resulting in numerous redundant computations and an inefficient use of limited energy resources, particularly for edge computing applications. To alleviate the high energy requirements associated with video frame processing, this paper presented similarity-aware CNNs that recognise similar feature pixels across frames and avoid computations on them. First, with a loss of less than 1% in recognition accuracy, a proposed similarity aware quantization technique increases the average number of unchanged feature pixels across frame pairs by up to 85%. Then, a proposed similarity-aware dataflow improves energy consumption by minimising redundant computations and memory accesses across frame pairs. According to simulation experiments, the proposed dataflow decreases the energy consumed by video frame processing by up to 30%.
Text
IEEE_TCAD_Final
- Accepted Manuscript
Text
Similarity-aware_CNN_for_Efficient_Video_Recognition_at_the_Edge
- Accepted Manuscript
Restricted to Repository staff only
Request a copy
More information
Accepted/In Press date: 7 December 2021
Published date: 20 December 2021
Additional Information:
Publisher Copyright:
IEEE
Keywords:
Computational modeling, Convolutional neural networks, Deep neural networks, Energy consumption, Memory management, Object Detection, Quantization, Quantization (signal), System-on-chip, Tensors, Video Recognition.
Identifiers
Local EPrints ID: 453181
URI: http://eprints.soton.ac.uk/id/eprint/453181
ISSN: 0278-0070
PURE UUID: 3a1f0db4-9337-43c0-a191-36a0b4184386
Catalogue record
Date deposited: 10 Jan 2022 18:01
Last modified: 17 Mar 2024 03:05
Export record
Altmetrics
Contributors
Author:
Mohammadamin Sabetsarvestani
Author:
Jonathon Hare
Author:
Bashir Al-Hashimi
Author:
Geoff Merrett
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics