Home

Held in conjunction with ICCV 2025 (Hawaii), Oct 19, 2025
Main Theme: Low-Power Computing for Embedded Vision
Organized by: Tse-Wei Chen, Branislav Kisacanin, Ahmed Nabil Belbachir, Marius Leordeanu

Description

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.

---

Program

BEST PAPER AWARD:
Adwait Chandorkar, Hasan Tercan, Tobias Meisen,
“Rethinking Backbone Design for Lightweight 3D Object Detection in LiDAR”

MOST INNOVATIVE PAPER AWARD:
Venkata Anirudh Puligandla, Vladimir Ceperic, Tihomir Knezevic,
“Scalable Optical Convolutional Neural Networks For Edge Applications”

Hawaii Time 10/19 (GMT-10)	Session	Speaker	Topic
9:00	Welcome notes	EVW Committee	Welcome notes
9:15	Invited Talk #1	Matteo Poggi	Stereo vision: from embedded systems to GPUs and back
9:45	Paper ID #2	Youcef Djenouri	LightPrune: Latency-Aware Structured Pruning for Efficient Deep Inference on Embedded Devices
10:00	Paper ID #3	Adwait Chandorkar	Rethinking Backbone Design for Lightweight 3D Object Detection in LiDAR
10:15	Break
10:30	Paper ID #4	Alberto Ancilotto	Contextual Convolutions for Scalable Forward-Only Learning on Tiny Devices
10:45	Paper ID #6	Sanmati Kamath	Implementation of Extremely Low Power Visual Perception Algorithm on Programmable Vision Accelerator: Examples of Challenges and Solutions
11:00	Paper ID #8	Venkata Anirudh Puligandla	Scalable Optical Convolutional Neural Networks For Edge Applications
11:15	Invited Talk #2	Yung-Hsiang Lu	Efficient Computer Vision for Edge Devices
11:45	Long Break
13:15	Invited Talk #3	Shingo Kagami	Super-Low-Latency Projectors with Embedded Warping Engines for Dynamic Interactive Systems
13:45	Invited Talk #4	Shao-Hua Sun	Imitation Learning with Diffusion Models
14:15	Paper ID #9	Daniel Airinei	Inside Knowledge: Graph-based Path Generation with Explainable Data Augmentation and Curriculum Learning for Visual Indoor Navigation
14:30	Paper ID #10	Sebastian Mocanu	Efficient Self-Supervised Neuro-Analytic Visual Servoing for Real-time Quadrotor Control
14:45	Poster Session and Break
15:45	Invited Talk #5	Jose Alvarez	System-Level Optimization for Real-Time Inference
16:15
16:15	Closing Remarks	EVW Committee

---

Important Dates

Paper submission: June 15, 2025 June 22, 2025 (23:59 GMT-7)
Demo abstract submission: June 15, 2025 June 22, 2025 (23:59 GMT-7)
Notification to the authors: July 5, 2025 July 12, 2025
Camera ready paper: August 4, 2025 August 18, 2025

Please refer to Submission page for details.
The accepted papers will be published in the proceeding, along with the ICCV main conference, indexed in EI Compendex.
OpenReview Submission website: https://openreview.net/group?id=thecvf.com/ICCV/2025/Workshop/EVW

---

Invited Talk #1

Invited Speaker: Prof. Matteo Poggi
Title: Stereo vision: from embedded systems to GPUs and back
Abstract: For decades, stereo matching algorithms have been extensively explored for deployment on embedded systems due to their relatively low computational and memory requirements. However, the advent of deep learning significantly shifted the landscape of stereo vision, scaling the computational complexity and memory demand while pursuing unprecedented accuracy. Indeed, state-of-the-art models often require high-end GPUs to run efficiently, which poses a substantial barrier for their integration into embedded or edge systems. With stereo becoming more reliable than ever and ready to be a milestone in higher-level applications of computer vision, it is time to focus back on real-time solutions suitable for deployment on embedded systems.
Biography: Matteo Poggi is a Tenure-Track Assistant Professore at the University of Bologna, where he completed the PhD in Computer Science and Engineering in 2018 under the supervision of Prof. Stefano Mattoccia. His research interests cover the broad area of 3D reconstruction from images and sensors, focusing in particular on depth estimation from one or multiple images and related tasks. On the same topics, he co-organized several workshops and tutorial at CVPR, ICCV and ECCV.

---

Invited Talk #2

Invited Speaker: Prof. Yung-Hsiang Lu
Title: Efficient Computer Vision for Edge Devices
Abstract: Since deep learning became popular a decade ago, computer vision has been adopted by a wide range of applications. Many applications must run on edge devices with limited resources (energy, time, memory capacity, etc). This speech will survey methods designed to improve efficiency of computer vision, including quantization, architecture search, and trade-off between accuracy and speed. A new architecture called Token Turing Machine is introduced. This architecture is based on vision transformer and creates two sets of tokens: process tokens and memory tokens;. Process tokens pass through encoder blocks; memory tokens can store and retrieve additional information from memory.  There are fewer process tokens than memory tokens and this architecture can reduce the inference time while maintaining its accuracy.
Biography: Yung-Hsiang Lu is a professor at the Elmore Family School of Electrical and Computer Engineering of Purdue University. In 2020-2022, he was the director of the John Martinson Engineering Entrepreneurial Center at Purdue University. He is a fellow of the IEEE, ACM Distinguished Scientist, ACM Distinguished Speaker, and Distinguished Visitor of the Computer Society. His research topics include efficient computer vision for embedded systems, cloud and mobile computing. He was the lead organizer of the IEEE Low-Power Computer Vision Challenge 2015-2025. He is an editor of the book “Low-Power Computer Vision Improve the Efficiency of Artificial Intelligence”, published by Chapman and Hall/CRC in 2022.

---

Invited Talk #3

Invited Speaker: Prof. Shingo Kagami
Title: Super-Low-Latency Projectors with Embedded Warping Engines for Dynamic Interactive Systems
Abstract: One of the key challenges in developing interactive systems using projectors is how to minimize display latency. A straightforward approach is to increase the video frame rate, but this significantly raises the overall system cost and complicates content production. Alternatively, by carefully exploiting the mechanism of the projection device, it is sometimes possible to reduce the perceived latency without increasing the frame rate. The speaker refers to such systems as super-low-latency displays. This talk will introduce implementations that enable this approach, with a particular focus on systems using Digital Micromirror Devices (DMDs). The talk will also present projection mapping systems designed for dynamic scenes, along with their applications.
Biography: Shingo Kagami received the B.E., M.E. and Ph.D. degrees in Mathematical Engineering and Information Physics from the University of Tokyo, Tokyo, Japan, in 1998, 2000, and 2003, respectively. He was a Research Fellow of Japan Science and Technology Corporation (JST) in 2003, and was a Research Associate of the University of Tokyo from 2003 to 2005. He joined Tohoku University, Sendai, Japan, in 2005, where he is currently a Professor of Unprecedented-scale Data Analytics Center and Graduate School of Information Sciences. His research interests include high-speed visual and sensory information processing and low-latency display systems for real-time real-world interaction and augmentation.

---

Invited Talk #4

Invited Speaker: Prof. Shao-Hua Sun
Title: Imitation Learning with Diffusion Models
Abstract: Imitation learning is a reinforcement learning paradigm in which agents learn to perform tasks by mimicking expert demonstrations, rather than relying on reward functions. It plays a central role in applications such as autonomous driving, robotics, and game AI. This talk will begin with a brief overview of imitation learning, introducing its core formulation and foundational algorithms, including Behavioral Cloning (BC) and Generative Adversarial Imitation Learning (GAIL). We will then transition to recent advances that integrate diffusion models into imitation learning frameworks. These approaches leverage the generative power of diffusion models to improve learning efficiency, generalizability, and data efficiency.
Biography: Shao-Hua Sun is an Assistant Professor in the Department of Electrical Engineering at National Taiwan University (NTU). He completed his Ph.D. in Computer Science at the University of Southern California (USC) and holds a B.S. in Electrical Engineering from NTU. Prof. Sun’s research interests include machine learning, robot learning, reinforcement learning, and program synthesis. His work has been presented at premier conferences across diverse fields, including machine learning (NeurIPS, ICML, ICLR), robot learning (CoRL), computer vision (CVPR, ECCV), and natural language processing and language modeling (EMNLP, COLM). He has organized tutorials at NeurIPS 2024 and ACML 2023, and workshops at ICML 20205 and RLC 2025.

---

Invited Talk #5

Invited Speaker: Dr. Jose Alvarez
Title of Talk: System-Level Optimization for Real-Time Inference
Abstract: Hardware resources are very limited when deploying deep neural networks for real-time applications such as autonomous driving. In these cases, the overall goal is to maximize the accuracy of neural network models while achieving the memory and latency constraints required for inference. A common approach to achieve this is to search for optimal architectures, train them and then use light optimization to fit the model to the desired hardware. In this talk, we present a learn large, inference light paradigm, where we start with very large models to maximize accuracy and then compress them very aggressively to meet the hardware limitations. We will walk through the pipeline highlighting the main modules including lossless model compression, pruning and distillation among others. As we will see, this paradigm also applies to modern transformer architectures.
Biography: Jose Alvarez leads a perception for autonomous driving research team. The team focuses on scaling up resource-constraint deep learning for autonomous driving, spanning scene understanding, 3D computer vision, self-supervised learning, data-efficient algorithms, and the efficiency of end-to-end perception models.
Before NVIDIA, Jose Alvarez held research positions at TRI, NICTA / CSIRO (Australia), and a postdoctoral research position at NYU under Prof. Yann LeCun.

---

Topics

• Agentic AI at the edge
• Embodied AI
• 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

---

Committee

---

Sponsors

---

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:
Dragos Costea, University Politehnica of Bucharest
Assia Belbachir, NORCE Norwegian Research Centre
Fahim Hasan Khan, California Polytechnic State University, San Luis Obispo
Florin Condrea, Siemens Corporate Research
Mihai Masala, The Institute of Mathematics of the Romanian Academy (IMAR)
Dongchao Wen, IEIT SYSTEMS Co., Ltd.
Alina Marcu, The National University of Science and Technology Politehnica Bucharest
Faycal Bensaali, University of Qatar
Sam Lerouxm Ghent University
Omkar Prabhunem Purdue University
Luca Bompani, University of Bologna
Linda M. Wills, Georgia Institute of Technology
Burak Ozer, Pekosoft LLC
Cevahir Cigla, ASELSAN
Natalia Jurado, Latent AI
Branislav Kisacanin, NVIDIA
Wei Tao, Canon Innovative Solution (Beijing) Co., Ltd.