Search Results for author: Ivan V. Bajić

Found 40 papers, 11 papers with code

Learned Compression of Encoding Distributions

1 code implementation18 Jun 2024 Mateen Ulhaq, Ivan V. Bajić

To address this issue, we propose a method that dynamically adapts the encoding distribution to match the latent data distribution for a specific input.

Decoder

Optimizing Split Points for Error-Resilient SplitFed Learning

no code implementations29 May 2024 Chamani Shiranthika, Parvaneh Saeedi, Ivan V. Bajić

Recent advancements in decentralized learning, such as Federated Learning (FL), Split Learning (SL), and Split Federated Learning (SplitFed), have expanded the potentials of machine learning.

Federated Learning Image Segmentation +1

Mutual Information Analysis in Multimodal Learning Systems

no code implementations21 May 2024 Hadi Hadizadeh, S. Faegheh Yeganli, Bahador Rashidi, Ivan V. Bajić

Taking advantage of the recent progress in entropy modeling and estimation, we develop a system called InfoMeter to estimate MI between modalities in a multimodal learning system.

3D Object Detection Autonomous Driving +1

Compressive Feature Selection for Remote Visual Multi-Task Inference

no code implementations15 May 2024 Saeed Ranjbar Alvar, Ivan V. Bajić

A key problem in applications such as feature compression for remote inference is determining how important each feature is for the task(s) performed by the model.

Feature Compression feature selection

Scalable Human-Machine Point Cloud Compression

1 code implementation19 Feb 2024 Mateen Ulhaq, Ivan V. Bajić

In this paper, we present a scalable codec for point-cloud data that is specialized for the machine task of classification, while also providing a mechanism for human viewing.

Learned Point Cloud Compression for Classification

1 code implementation11 Aug 2023 Mateen Ulhaq, Ivan V. Bajić

Our codec demonstrates the potential of specialized codecs for machine analysis of point clouds, and provides a basis for extension to more complex tasks and datasets in the future.

Classification object-detection +1

SplitFed resilience to packet loss: Where to split, that is the question

no code implementations25 Jul 2023 Chamani Shiranthika, Zahra Hafezi Kafshgari, Parvaneh Saeedi, Ivan V. Bajić

Decentralized machine learning has broadened its scope recently with the invention of Federated Learning (FL), Split Learning (SL), and their hybrids like Split Federated Learning (SplitFed or SFL).

Federated Learning

Metaverse: A Young Gamer's Perspective

no code implementations19 Jul 2023 Ivan V. Bajić, Teo Saeedi-Bajić, Kai Saeedi-Bajić

When developing technologies for the Metaverse, it is important to understand the needs and requirements of end users.

Learned Scalable Video Coding For Humans and Machines

1 code implementation18 Jul 2023 Hadi Hadizadeh, Ivan V. Bajić

Video coding has traditionally been developed to support services such as video streaming, videoconferencing, digital TV, and so on.

Object Detection Video Compression

Base Layer Efficiency in Scalable Human-Machine Coding

no code implementations5 Jul 2023 Yalda Foroutan, Alon Harell, Anderson de Andrade, Ivan V. Bajić

A basic premise in scalable human-machine coding is that the base layer is intended for automated machine analysis and is therefore more compressible than the same content would be for human viewing.

Instance Segmentation object-detection +2

Grad-FEC: Unequal Loss Protection of Deep Features in Collaborative Intelligence

no code implementations4 Jul 2023 Korcan Uyanik, S. Faegheh Yeganli, Ivan V. Bajić

Collaborative intelligence (CI) involves dividing an artificial intelligence (AI) model into two parts: front-end, to be deployed on an edge device, and back-end, to be deployed in the cloud.

Feature Importance

Adversarial Attacks and Defenses on 3D Point Cloud Classification: A Survey

no code implementations1 Jul 2023 Hanieh Naderi, Ivan V. Bajić

To encourage future research, this survey summarizes the current progress on adversarial attack and defense techniques on point cloud classification. This paper first introduces the principles and characteristics of adversarial attacks and summarizes and analyzes adversarial example generation methods in recent years.

3D Point Cloud Classification Adversarial Attack +2

Quality-Adaptive Split-Federated Learning for Segmenting Medical Images with Inaccurate Annotations

no code implementations28 Apr 2023 Zahra Hafezi Kafshgari, Chamani Shiranthika, Parvaneh Saeedi, Ivan V. Bajić

SplitFed Learning, a combination of Federated and Split Learning (FL and SL), is one of the most recent developments in the decentralized machine learning domain.

Federated Learning Image Segmentation +2

Multi-Task Learning for Screen Content Image Coding

1 code implementation3 Feb 2023 Rashid Zamanshoar Heris, Ivan V. Bajić

With the rise of remote work and collaboration, compression of screen content images (SCI) is becoming increasingly important.

Multi-Task Learning Segmentation

LCCM-VC: Learned Conditional Coding Modes for Video Compression

1 code implementation28 Oct 2022 Hadi Hadizadeh, Ivan V. Bajić

End-to-end learning-based video compression has made steady progress over the last several years.

Video Compression

Privacy-Preserving Feature Coding for Machines

no code implementations3 Oct 2022 Bardia Azizian, Ivan V. Bajić

We present a novel method to create a privacy-preserving latent representation of an image that could be used by a downstream machine vision model.

Privacy Preserving

Scalable Video Coding for Humans and Machines

no code implementations4 Aug 2022 Hyomin Choi, Ivan V. Bajić

Video content is watched not only by humans, but increasingly also by machines.

MS-SSIM Object +3

Adversarial Attacks on Human Vision

no code implementations3 Jun 2022 Victor A. Mateescu, Ivan V. Bajić

This article presents an introduction to visual attention retargeting, its connection to visual saliency, the challenges associated with it, and ideas for how it can be approached.

Joint Image Compression and Denoising via Latent-Space Scalability

no code implementations4 May 2022 Saeed Ranjbar Alvar, Mateen Ulhaq, Hyomin Choi, Ivan V. Bajić

In this paper, we present a learning-based image compression framework where image denoising and compression are performed jointly.

Image Compression Image Denoising +1

License Plate Privacy in Collaborative Visual Analysis of Traffic Scenes

no code implementations3 May 2022 Saeed Ranjbar Alvar, Korcan Uyanik, Ivan V. Bajić

Traffic scene analysis is important for emerging technologies such as smart traffic management and autonomous vehicles.

Autonomous Vehicles Management

Does Video Compression Impact Tracking Accuracy?

no code implementations2 Feb 2022 Takehiro Tanaka, Alon Harell, Ivan V. Bajić

Everyone "knows" that compressing a video will degrade the accuracy of object tracking.

Multiple Object Tracking Object +2

Practical Noise Simulation for RGB Images

no code implementations30 Jan 2022 Saeed Ranjbar Alvar, Ivan V. Bajić

This document describes a noise generator that simulates realistic noise found in smartphone cameras.

Image Denoising

SFU-HW-Tracks-v1: Object Tracking Dataset on Raw Video Sequences

no code implementations30 Dec 2021 Takehiro Tanaka, Hyomin Choi, Ivan V. Bajić

We present a dataset that contains object annotations with unique object identities (IDs) for the High Efficiency Video Coding (HEVC) v1 Common Test Conditions (CTC) sequences.

Object Object Tracking +1

DFTS2: Simulating Deep Feature Transmission Over Packet Loss Channels

2 code implementations1 Dec 2021 Ashiv Dhondea, Robert A. Cohen, Ivan V. Bajić

In edge-cloud collaborative intelligence (CI), an unreliable transmission channel exists in the information path of the AI model performing the inference.

Image Classification Packet Loss Concealment

CALTeC: Content-Adaptive Linear Tensor Completion for Collaborative Intelligence

1 code implementation10 Jun 2021 Ashiv Dhondea, Robert A. Cohen, Ivan V. Bajić

In collaborative intelligence, an artificial intelligence (AI) model is typically split between an edge device and the cloud.

Error Resilient Collaborative Intelligence via Low-Rank Tensor Completion

no code implementations20 May 2021 Lior Bragilevsky, Ivan V. Bajić

The communication channel between the edge and cloud is imperfect, which will result in missing data in the deep feature tensor received at the cloud side.

Lightweight Compression of Intermediate Neural Network Features for Collaborative Intelligence

no code implementations15 May 2021 Robert A. Cohen, Hyomin Choi, Ivan V. Bajić

In collaborative intelligence applications, part of a deep neural network (DNN) is deployed on a lightweight device such as a mobile phone or edge device, and the remaining portion of the DNN is processed where more computing resources are available, such as in the cloud.

object-detection Object Detection +1

Lightweight compression of neural network feature tensors for collaborative intelligence

no code implementations12 May 2021 Robert A. Cohen, Hyomin Choi, Ivan V. Bajić

In collaborative intelligence applications, part of a deep neural network (DNN) is deployed on a relatively low-complexity device such as a mobile phone or edge device, and the remainder of the DNN is processed where more computing resources are available, such as in the cloud.

object-detection Object Detection

Swimmer Stroke Rate Estimation From Overhead Race Video

no code implementations25 Apr 2021 Timothy Woinoski, Ivan V. Bajić

In this work, we propose a swimming analytics system for automatically determining swimmer stroke rates from overhead race video (ORV).

Collaborative Intelligence: Challenges and Opportunities

no code implementations13 Feb 2021 Ivan V. Bajić, Weisi Lin, Yonghong Tian

This paper presents an overview of the emerging area of collaborative intelligence (CI).

Feature Compression

Analysis of Latent-Space Motion for Collaborative Intelligence

no code implementations8 Feb 2021 Mateen Ulhaq, Ivan V. Bajić

When the input to a deep neural network (DNN) is a video signal, a sequence of feature tensors is produced at the intermediate layers of the model.

Optical Flow Estimation

Latent-Space Inpainting for Packet Loss Concealment in Collaborative Object Detection

no code implementations30 Jan 2021 Ivan V. Bajić

Edge devices, such as cameras and mobile units, are increasingly capable of performing sophisticated computation in addition to their traditional roles in sensing and communicating signals.

Image Inpainting Object +3

Analysis of Information Flow Through U-Nets

1 code implementation21 Jan 2021 Suemin Lee, Ivan V. Bajić

Deep Neural Networks (DNNs) have become ubiquitous in medical image processing and analysis.

Image Segmentation Semantic Segmentation

Pareto-Optimal Bit Allocation for Collaborative Intelligence

no code implementations25 Sep 2020 Saeed Ranjbar Alvar, Ivan V. Bajić

Moreover, we provide analytical characterization of the full Pareto set for 2-stream k-task systems, and bounds on the Pareto set for 3-stream 2-task systems.

Bit Allocation for Multi-Task Collaborative Intelligence

no code implementations14 Feb 2020 Saeed Ranjbar Alvar, Ivan V. Bajić

In CI, a deep neural network is split between the mobile device and the cloud.

Shared Mobile-Cloud Inference for Collaborative Intelligence

no code implementations1 Feb 2020 Mateen Ulhaq, Ivan V. Bajić

Partial inference is performed on the mobile in order to reduce the dimensionality of the input data and arrive at a compact feature tensor, which is a latent space representation of the input signal.

Wavenilm: A causal neural network for power disaggregation from the complex power signal

1 code implementation23 Feb 2019 Alon Harell, Stephen Makonin, Ivan V. Bajić

Non-intrusive load monitoring (NILM) helps meet energy conservation goals by estimating individual appliance power usage from a single aggregate measurement.

Non-Intrusive Load Monitoring

Cannot find the paper you are looking for? You can Submit a new open access paper.