Held in conjunction with CVPR 2022 (New Orleans), June 20 (Full Day), 2022
Main Theme: Edge Artificial Intelligence
Organized by: Marius Leordeanu, Ahmed Nabil Belbachir, Tse-Wei Chen


Embedded vision is an active field of research, bringing together efficient learning models with fast computer vision and pattern recognition algorithms, to tackle many areas of robotics and intelligent systems that are enjoying an impressive growth today. Such strong impact comes with many challenges that stem from the difficulty of understanding complex visual scenes under the tight computational constraints required by real-time solutions on embedded devices. The Embedded Vision Workshop will provide a venue for discussing these challenges by bringing together researchers and practitioners from the different fields outlined above. Such a topic is directly aligned with the Topics of interest of the CVPR community.


Important: EVW2022 will be a virtual-only event. To attend the workshop, please access the link on the CVPR virtual platform.

New Orleans Time
6/20 (GMT-5)
8:45Welcome notesEVW CommitteeWelcome notes
9:00Invited Talk #1Osamu NomuraAnalog Neuromorphic Hardware for Energy-efficient AI Computing
9:30Invited Talk #2Rongrong JiCompression and Acceleration of Deep Neural Networks
10:15Paper ID #3Swarnava DeySymDNN: Simple & Effective Adversarial Robustness for Embedded Systems
10:30Paper ID #4Indhumathi KandaswamyReal-time Hyper-Dimensional Reconfiguration at the Edge using Hardware Accelerators
10:45Paper ID #6Jordan H ShipardDoes Interference Exist When Training a Once-For-All Network?
11:00Paper ID #10Yanan LiuOn-Sensor Binarized Fully Convolutional Neural Network for Localisation and Coarse Segmentation
11:15Demo ID #11Mohammad Javad ShafieeLightDefectNet: A Highly Compact Deep Anti-Aliased Attention Condenser Neural Network Architecture for Light Guide Plate Surface Defect Detection
11:30Invited Talk #3Yasutomo KawanishiButukusa-2: A prototype Robot for Human Assistance
12:00Long Break
13:00Invited Talk #4Robert JenssenComputer Vision for Power Line Monitoring
13:30Invited Talk #5Matthias GrundmannLive Perception for Mobile and Web 2
14:15Paper ID #13Mustafa AyazogluEfficient Multi-Purpose Cross-Attention Based Image Alignment Block for Edge Devices
14:30Paper ID #14Mohamed R IbrahimImageSig: A signature transform for ultra-lightweight image recognition
14:45Paper ID #19Saeejith NairMAPLE-Edge: A Runtime Latency Predictor for Edge Devices
15:00Paper ID #20Zejiang HouMulti-Dimensional Vision Transformer Compression via Dependency Guided Gaussian Process Search
15:15Demo ID #17Saad AbbasiMAPLE-X: Latency Prediction with Explicit Microprocessor Prior Knowledge
15:30Closing RemarksEVW Committee

Invited Speakers

Osamu Nomura
Title of Talk: Analog Neuromorphic Hardware for Energy-efficient AI Computing
Abstract: Artificial neural networks (ANNs) have shown excellent performance on various tasks, including image recognition, speech recognition, anomaly detection, and so forth. Although ANNs can achieve high accuracies on the tasks, their high energy consumption is a major challenge due to a large number of calculations. To implement AI functions on edge devices, operations at low-power consumption are required. For these requirements, studies on neuromorphic hardware have been reported in recent years. Neuromorphic hardware mimics neuro-biological models and has the potential to achieve high energy efficiency.
In this talk, I will introduce basic concepts for neuromorphic computing, and present some of our group’s achievements and plans on neuromorphic models and hardware implementations with analog circuit techniques.
Biography: Osamu Nomura received the B.E. degree in mechanical engineering from Yokohama National University, Yokohama, Japan, the Dr.Eng. degree from Kyushu Institute of Technology, Kitakyushu, Japan, and the Dr.Sci. degree from Tokyo Institute of Technology, Tokyo, Japan, in 1993, 2006, and 2020, respectively. From 1993 to 2020, he was an engineer at Canon Inc. From 2020 to 2021, he was an Associate Professor of the Center for Innovative Integrated Electronic Systems, Tohoku University, Sendai, Japan. Since 2021 he has been a Specially Appointed Professor of the Graduate School of Life Science and Systems Engineering, Kyushu Institute of Technology (Kyutech), Kitakyushu, Japan. He is a member of Research Center for Neuromorphic AI Hardware, Kyutech. His research interests include models of brain function, Edge AI implementation, and new functional nanodevices. He was selected as the Best Paper at IEEE MWSCAS 2022.

Rongrong Ji
Title of Talk: Compression and Acceleration of Deep Neural Networks
Abstract: Deep neural networks have made remarkable achievements in artificial intelligence applications such as image understanding, speech recognition, and natural language processing, and have become one of research focus of Artificial Intelligence. However, with the continuous improvement of network performance, the depth and breadth of the networks are also growing explosively, which greatly increases the parameters and computational complexity of deep learning models. How to compress and accelerate these large neural network models has become a research hot spot in both academia and industry. Aiming at the acceleration and redundancy removal of deep neural networks, this talk briefly introduces the existing acceleration and compression methods and covers some of the recent work and achievements of Professor Rongrong Ji’s research group.
Biography: Prof. Rongrong Ji is a distinguished professor at Xiamen University, a recipient of the National Natural Science Fund for Distinguished Young Scholars. His research falls in the field of computer vision, multimedia analysis, and machine learning. He has published 100+ papers in ACM/IEEE Transactions, including TPAMI and IJCV, as well as top-tier international conferences, such as CVPR and NeurIPS. His publications have got over 10K citations in Google Scholar. He was the recipient of the first prize of technology invention of the ministry of education in 2016, the first prize of the Fujian provincial science and technology award in 2018, science and technology award for youth of Fujian province in 2019. He has served as the area chair of top-tier international conferences such as IEEE CVPR and ACM Multimedia. He is also the Vice Director of Academic Working Committee of Chinese Society of Image and Graphics, and a member of the Artificial Intelligence Professional Construction Advisory Committee of the Electronic Information Education Commission of the Ministry of Education.

Yasutomo Kawanishi
Title of Talk: Butukusa-2: A prototype robot for human assistance
Abstract: The Guardian Robot Project (GRP), promoted by RIKEN, aims to develop autonomous mobile robots that can be familiar to people and make people feel the “heart.” The robots are expected to operate proactively and casually assist people, especially the elderly who live alone. Butukusa-2, the current prototype, has a function that assists people’s memory. The robot patrols a room and memorizes objects’ names and their locations in the room. And then, it answers their locations when asked. In this talk, I will introduce the implementation of the robot and the environment recognition technologies implemented on the robot.
Biography: Yasutomo Kawanishi received the B.Eng. degree in engineering and the M.Inf. and Ph.D. degrees in informatics from Kyoto University, Japan, in 2006, 2008, and 2012, respectively. He became a Postdoctoral Fellow with Kyoto University, in 2012. In 2014, he moved to Nagoya University, Japan, as a Designated Assistant Professor. In 2015, he became an Assistant Professor and a Lecturer, in 2020. Since 2021, he has been the Team Leader of the Multimodal Data Recognition Research Team, RIKEN Guardian Robot Project. He is a fellow of the 4th Intercontinental Academia (ICA4). His main research interests include robot vision for environmental understanding and computer vision for human understanding, especially pedestrian detection, tracking, retrieval, and recognition. He is a member of IEEE, IIEEJ and IEICE. He received the Best Paper Award from SPC2009 and the Young Researcher Award from the IEEE ITS Society Nagoya Chapter.

Robert Jenssen
Title of Talk: Computer vision for power line monitoring
Abstract: This talk will describe development of deep learning-based computer vision for monitoring of power lines from images taken from helicopters and drones. In particular, the talk will describe a novel neural network for learning to predict line structures. The line structures correspond in this application to the power lines and the long-term task is to aid autonomous navigation of the drones. The talk builds on deep learning research conducted within the scientific program of the centre for research-based innovation “Visual Intelligence” in Norway: http://visual-intelligence.no
Biography: Robert Jenssen (Senior Member, IEEE) received the Ph.D. (Dr. Scient.) degree in electrical engineering from the University of Tromsø, Tromsø, Norway, in 2005.,He is a Professor and the Head of the UiT Machine Learning Group, UiT The Arctic University of Norway.,Dr. Jenssen received the 2005 IEEE ICASSP Outstanding Student Paper Award, the 2007 UiT Young Investigator Award, and the 2013 IEEE Geoscience and Remote Sensing Society Letters Best Paper Award. He is the General Chair of the annual Northern Lights Deep Learning (NLDL) Workshop. He serves on the IEEE Technical Committee on Machine Learning for Signal Processing, he is on the International Association for Pattern Recognition (IAPR) Governing Board, and he is an Associate Editor for the journal Pattern Recognition.

Matthias Grundmann
Title of Talk: On-device ML solutions for Mobile and Web
Abstract: In this talk, I will present several on-device Machine Learning (ML) solutions for mobile and web that are powering a wide range of impactful Google Products. On-device ML has major benefits enabling low-latency, offline and privacy-preserving approaches. However, to ship these solutions in production, we need to overcome substantial technical challenges to deliver on-device ML in real-time and with low-latency. Once solved, our solutions power applications like background replacement and light adjustment in Google Meet, AR effects in YouTube and Duo, gesture controls of devices and view-finder tracking for Google Lens and Translate. 
In this talk, I will cover some of the core-recipes behind Google’s on-device ML solutions, from model design over enabling ML solutions infrastructure (MediaPipe) to on-device ML inference acceleration. In particular we will be covering video segmentation, face meshes and iris tracking, hand tracking for gesture control and body tracking to power 3D avatars. The covered solutions are also available to the research and developer community via MediaPipe, —an open source cross platform framework for building customizable ML pipelines for mobile, web, desktop and python.
Biography: Matthias Grundmann is a Director of Research at Google leading a team of ~70 Applied ML, Software Engineers and Performance Experts with focus on on-device Machine Learning solutions. His team develops high-quality, cross-platform ML solutions (MediaPipe) powered by cutting-edge, accelerated ML inference for mobile and web. 
His team productionized many on-device solutions ranging from video segmentation for Google Meet and YouTube, over 2D object and calibration-free 6 DOF camera tracking, to computational video solutions powering Light Adjustment in Google Meet, Motion Photos on Pixel and Live Photo stabilization in Google Photos.
Among the wide portfolio of Applied ML solutions his team develops are holistic methods for hand, body 3D object, and high-fidelity facial geometry tracking. His team has advanced on-device ML technology across Google delivering sparsity powering Google Meet, quantization and on-device CPU and GPU inference.
Matthias received his Ph.D. from the Georgia Institute of Technology in 2013 for his work on Computational Video with focus on Video Stabilization and Rolling Shutter removal for YouTube. His work on Rolling Shutter removal won the best paper award at ICCP, 2012. He was the recipient of the 2011 Ph.D. Google Fellowship in Computer Vision.

Important Dates

Paper submission: March 18, 2022
Demo abstract submission: March 18, 2022
Notification to the authors: April 10, 2022
Camera ready paper: April 15, 2022 April 20, 2022

Please refer to Submission page for details.
(Supplementary material can be uploaded until March 18, 2022.)


  • Lightweight and efficient computer vision algorithms for embedded systems
  • Hardware dedicated to embedded vision systems (GPUs, FPGAs, DSPs, etc.)
  • Software platforms for embedded vision systems
  • Neuromorphic computing
  • Applications of embedded vision systems in general domains: UAVs (industrial, mobile and consumer), Advanced assistance systems and autonomous navigation frameworks, Augmented and Virtual Reality, Robotics.
  • New trends and challenges in embedded visual processing
  • Analysis of vision problems specific to embedded systems
  • Analysis of embedded systems issues specific to computer vision
  • Biologically-inspired vision and embedded systems
  • Hardware and software enhancements that impact vision applications
  • Performance metrics for evaluating embedded systems
  • Hybrid embedded systems combining vision and other sensor modalities
  • Embedded vision systems applied to new domains


General Chair:
Marius Leordeanu, University Politehnica Bucharest (Romania)

General Chair:
Ahmed Nabil Belbachir, NORCE Norwegian Research Centre (Norway)

Publication Chair:
Tse-Wei Chen, Canon Inc. (Japan)

Steering Committee:
Marilyn Claire Wolf, University of Nebraska-Lincoln
Martin Humenberger, NAVER LABS Europe
Roland Brockers, Jet Propulsion Laboratory
Swarup Medasani, MathWorks
Stefano Mattoccia, University of Bologna
Jagadeesh Sankaran, Nvidia
Goksel Dedeoglu, Perceptonic
Margrit Gelautz, Vienna University of Technology
Branislav Kisacanin, Nvidia
Sek Chai, Latent AI
Zoran Nikolic, Nvidia
Ravi Satzoda, Nauto
Stephan Weiss, University of Klagenfurt

Program Committee:
Alina Marcu, University Politehnica of Bucharest
Antonio Haro, eBay
Burak Ozer, Verificon Corporation
Daniel Steininger, AIT Austrian Institute of Technology
Dongchao Wen, State Key Laboratory of High-end Server & Storage Technology
Dragos Costea, University Politehnica of Bucharest
Faycal Bensaali, Qatar University
Florin Condrea, Institute of Mathematics of the Romanian Academy
Linda Wills, Georgia Institute of Technology
Martin Kampel, Vienna University of Technology, Computer Vision Lab
Matteo Poggi, University of Bologna
Matthias Schörghuber, AIT Austrian Institute of Technology
Mihai Cristian Pîrvu, University Politehnica of Bucharest
Nabil Belbachir, NORCE Norwegian Research Centre AS
Tse-Wei Chen, Canon Inc.

%d bloggers like this: