Computer Vision

Computer Vision

Our research interests include learning with limited labels, cross-domain, self-supervised and multi-modal learning, and modern model architectures. We focus on innovative state-of-the-art research that makes a difference.

Read more about Computer Vision at IBM Research - Haifa

Read more about Medical Imaging Analytics and Solutions at IBM Research - Haifa

 

Our research interests include learning with limited labels, cross-domain, self-supervised and multi-modal learning, and modern model architectures. We focus on innovative state-of-the-art research that makes a difference.

Read more about Computer Vision at IBM Research - Haifa
Read more about Medical Imaging Analytics and Solutions at IBM Research - Haifa

Collaboration & Publications

Publications that were published during 2021

Title Author Year Conference/Journal Area  
Fine-grained Angular Contrastive Learning with Coarse Labels Guy Bukchin, Eli Schwartz, Kate Saenko, Ori Shahar, Rogerio Feris, Raja Giryes, Leonid Karlinsky 2021 CVPR Self-Supervised Learning, Few-Shot Learning, Few-Shot Classification Link
StarNet: towards weakly supervised few-shot detection and explainable few-shot classification Leonid Karlinsky*, Joseph Shtok*, Amit Alfassy*, Moshe Lichtenstein*, Sivan Harary, Eli Schwartz, Sivan Doveh, Prasanna Sattigeri, Rogerio Feris, Alexander Bronstein, Raja Giryes 2021 AAAI Few-Shot Learning, Few-Shot Classification, Few-Shot Detection Link
MetAdapt: Meta-Learned Task-Adaptive Architecture for Few-Shot Classification Sivan Doveh, Eli Schwartz, Chao Xue, Rogerio Feris, Alex Bronstein, Raja Giryes, Leonid Karlinsky 2021 Pattern Recognition Letters Few-Shot Learning, Few-Shot Classification Link
Dynamic Distillation Network for Cross-Domain Few-Shot Recognition with Unlabeled Data A Islam, CF Chen, R Panda, L Karlinsky, R Feris, RJ Radke 2021 NeurIPS Few-Shot Learning, Few-Shot Classification, Cross-domain Few-Shot Learning Link
Detector-Free Weakly Supervised Grounding by Separation Assaf Arbelle, Sivan Doveh, Amit Alfassy, Joseph Shtok, Guy Lev, Eli Schwartz, Hilde Kuehne, Hila Barak Levi, Prasanna Sattigeri, Rameswar Panda, Chun-Fu Chen, Alex Bronstein, Kate Saenko, Shimon Ullman, Raja Giryes, Rogerio Feris, Leonid Karlinsky 2021 ICCV Text grounding in images Link
A Broad Study on the Transferability of Visual Representations with Contrastive Learning Ashraful Islam, Chun-Fu Chen, Rameswar Panda, Leonid Karlinsky, Richard Radke, Rogerio Feris 2021 ICCV Transfer Learning Link
Adafuse: Adaptive temporal fusion network for efficient action recognition Yue Meng, Rameswar Panda, Chung-Ching Lin, Prasanna Sattigeri, Leonid Karlinsky, Kate Saenko, Aude Oliva, Rogerio Feris 2021 ICLR Action Recognition, Efficient AI Link
CHARTER: heatmap-based multi-type chart data extraction Joseph Shtok, Sivan Harary, Ophir Azulai, Adi Raz Goldfarb, Assaf Arbelle, Leonid Karlinsky 2021 KDD DIW Charts analysis  
Noise estimation using density estimation for self-supervised multimodal learning E Amrani, R Ben-Ari, D Rotman, A Bronstein 2021 AAAI Self-Supervised multimodal learning Link

More publications

Title Author Year Conference/Journal Area  
Ar-net: Adaptive frame resolution for efficient action recognition Authors Yue Meng, Chung-Ching Lin, Rameswar Panda, Prasanna Sattigeri, Leonid Karlinsky, Aude Oliva, Kate Saenko, Rogerio Feris 2020 ECCV Action Recognition, Efficient AI Link
OnlineAugment: Online data augmentation with less domain knowledge Zhiqiang Tang, Yunhe Gao, Leonid Karlinsky, Prasanna Sattigeri, Rogerio Feris, Dimitris Metaxas 2020 ECCV Transfer Learning Link
TAFSSL: Task-Adaptive Feature Sub-Space Learning for few-shot classification Moshe Lichtenstein, Prasanna Sattigeri, Rogerio Feris, Raja Giryes, Leonid Karlinsky 2020 ECCV Few-Shot Learning, Few-Shot Classification, Semi-Supervised Learning, Transductive Learning Link
A broader study of cross-domain few-shot learning Yunhui Guo, Noel C Codella, Leonid Karlinsky, James V Codella, John R Smith, Kate Saenko, Tajana Rosing, Rogerio Feris 2020 ECCV Few-Shot Learning, Few-Shot Classification, Cross-domain Few-Shot Learning Link
A Maximal Correlation Approach to Imposing Fairness in Machine Learning Joshua Lee, Yuheng Bu, Prasanna Sattigeri, Rameswar Panda, Gregory Wornell, Leonid Karlinsky, Rogerio Feris 2020 arXiv preprint arXiv:2012.15259 Fairness Link
Baby steps towards few-shot learning with multiple semantics Eli Schwartz*, Leonid Karlinsky*, Rogerio Feris, Raja Giryes, Alex M Bronstein 2019 arXiv preprint arXiv:1906.01905 Few-Shot Learning, Few-Shot Classification, Multimodal Learning Link
A CNN based method for automatic mass detection and classification in mammograms Ayelet Akselrod-Ballin, Leonid Karlinsky, Sharon Alpert, Sharbell Hashoul, Rami Ben-Ari, Ella Barkan 2019 Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization (CMBBE) Medical Vision  
Repmet: Representative-based metric learning for classification and few-shot object detection Leonid Karlinsky*, Joseph Shtok*, Sivan Harary*, Eli Schwartz*, Amit Aides, Rogerio Feris, Raja Giryes, Alex M Bronstein 2019 CVPR Few-Shot Learning, Few-Shot Detection Link
Laso: Label-set operations networks for multi-label few-shot learning Amit Alfassy*, Leonid Karlinsky*, Amit Aides, Joseph Shtok, Sivan Harary, Rogerio Feris, Raja Giryes, Alex M Bronstein 2019 CVPR Transfer Learning, Few-Shot Learning, Few-Shot Classification Link
Co-regularized alignment for unsupervised domain adaptation Abhishek Kumar, Prasanna Sattigeri, Kahini Wadhawan, Leonid Karlinsky, Rogerio Feris, Bill Freeman, Gregory Wornell 2018 NeurIPS Domain Adaptation Link
Delta-encoder: an effective sample synthesis method for few-shot object recognition Eli Schwartz*, Leonid Karlinsky*, Joseph Shtok, Sivan Harary, Mattias Marder, Abhishek Kumar, Rogerio Feris, Raja Giryes, Alex Bronstein 2018 NeurIPS Few-Shot Learning, Few-Shot Classification Link
Domain specific convolutional neural nets for detection of architectural distortion in mammograms Rami Ben-Ari, Ayelet Akselrod-Ballin, Leonid Karlinsky, Sharbell Hashoul 2017 ISBI Medical Vision Link
Fine-grained recognition of thousands of object categories with single-example training Leonid Karlinsky, Joseph Shtok, Yochay Tzur, Asaf Tzadok 2017 CVPR Few-Shot Learning, Few-Shot Classification, Few-Shot Detection Link
Hybrid remote expert-an emerging pattern of industrial remote support Ethan Hadar, Joseph Shtok, Benjamin Cohen, Yochay Tzur, Leonid Karlinsky 2017 CAISE 2017 AR Link
Deep learning for automatic detection of abnormal findings in breast mammography Ayelet Akselrod-Ballin, Leonid Karlinsky, Alon Hazan, Ran Bakalo, Ami Ben Horesh, Yoel Shoshan, Ella Barkan 2017 Deep learning in medical image analysis and multimodal learning for clinical decision support Medical Vision  
A region based convolutional network for tumor detection and classification in breast mammography Ayelet Akselrod-Ballin, Leonid Karlinsky, Sharon Alpert, Sharbell Hasoul, Rami Ben-Ari, Ella Barkan 2016 Deep learning and data labeling for medical applications (DLDL) Medical Vision  
Deep Learning and Data Labeling for Medical Applications A Akselrod-Ballin, L Karlinsky, S Alpert, S Hasoul, R Ben-Ari, E Barkan, G Carneiro 2016 Int. Work. Large-Scale Annot. Biomed. Data Expert Label Synth Medical Vision  

Workshops & Tutorials

Women in AI Israel (2019, 2020, 2021) 

A community-driven initiative aimed at increasing the number of women working in AI and introducing the AI community to female researchers & practitioners.

CVPR Workshop (2020)

VL3: Visual Learning with Limited Labels.

ICCV Tutorial (2019)

VISUAL LEARNING WITH LIMITED LABELED DATA 

Open Source Activities

LASO

This repository contains the implementation of "LaSO: Label-Set Operations networks for multi-label few-shot learning" by Alfassy et al. It was posted on arxiv in Feb 2019 and will be presented in CVPR 2019.

RepMet

RepMet few-shot detection engine and an Imagenet-LOC detection benchmark

DeltaEncoder

Implementation for the paper "Delta-encoder: an effective sample synthesis method for few-shot object recognition" from NeurIPS 2018

Cross-Domain Few-Shot Learning (CD-FSL) Benchmark

Cross-Domain Few-Shot Learning (CD-FSL) Benchmark

Multi-scale, Multi-object Star Model detection algorithm

Multi-scale multi-object star model Matlab code for single shot object detection & recognition. Trained using one or few examples per object, can support up to few thousands of objects at once.

ANCOR

Fine-grained Angular Contrastive Learning with Coarse Labels (CVPR 2021 Oral)

TAFSSL

Task-Adaptive Feature Sub-Space Learning for few-shot classification

AR-Net

Adaptive Resolution Network for Efficient Video Understanding

Online-augment

Official OnlineAugment implementation in PyTorch

AdaFuse

Adaptive Temporal Fusion Network for Efficient Action Recognition (ICLR 2021)

Awards

Eli Schwartz (Tel Aviv University):
IBM PhD Fellowship (2020)


Eli Schwartz has received the IBM PhD Fellowship, a competitive award given to PhD students who have demonstrated academic excellence as well as provided innovative, exceptional research proposals.

IBM Research Outstanding Innovation Award (OIA) 2020

Contribution to science in the Learning with Less Labels (LwLL) domain – data limited learning for vision and beyond” science accomplishment for the year 2020. This accomplishment is a cross-lab effort. The total contributions included: 16 papers with over 400 citations, 17 patents, and 3 successful customer engagements, as well as numerous academy collaborations.   

IBM Research Dynamic Neural Networks for Efficient AI (OIA) 2020

(Non-invited) Venturebeat article published by an independent journalist and describing Leonid’s StarNet AAAI 2021 paper. Through this article IBM’s explainable AI toolkit gets promoted and compared as equal to similar efforts by Facebook, Microsoft, and Google.

 

Task Adaptive Feature Sub-Space Learning (TAFSSL)

In this work, we explored ways of how unlabeled data of different kinds can be leveraged to significantly boost few-shot learning performance.

Our paper TAFSSL (listed as ICA + MSP) is #1 for 1-shot (were #1 in 2020)

We are #3 in 1-shot (were #1 in 2020)

Talks

Leonid Karlinsky

Explainable, Adaptive, and Cross-Domain Few-Shot Learning

Invited talk for the 2d3d.ai meetup 2021

Watch Part 01
Watch Part 02

Eli Schwartz

Small-Data in the Big-Data Era

Invited talk for the Deep Corona Academy - Alibaba meetup

Watch

Guy Bbukchin (Intern)

Fine-grained Angular Contrastive Learning with Coarse Labels

Oral Session for CVPRi 2020 (hebrew)

Watch

 

More talks

Sivan Doveh - Meta-learning: few-shot learning and neural architecture search, Invited talk for the Deep Corona Academy - Alibaba meetup

Eli Schwartz - Small-Data in the Big-Data Era, Invited Talk, The 6th Ambassadors' Summit, Tel Aviv University

Leonid Karlinsky - ECCVi 2020 (hebrew)

Amit Alfassy - LaSO: Label-Set Operations networks for multi-label few-shot learning, Oral Session for CVPR 2019.

Leonid Karlinsky - IMVC 2019

Eli Schwartz - Few-shot learning with multiple semantics, CVPR 2019 Language And Vision Workshop

Leonid Karlinsky - Data Science Summit 2018

Eli Schwartz - Presenting our NeurIPS 2018 paper

Eli Schwartz - Few-shot learning, at the Geometric Computing Lab, Stanford University, USA, October 2018

Let's talk

Want to collaborate? We're always happy to talk. Feel free to get in touch.

 


Computer vision Area Focal Point

Working together
with IBM Research in Haifa

Our published papers

Our catalog of recent publications authored by IBM researchers, in collaboration with the global research community.


Explore

Open positions and internships

Join some of the best scientists from around the world, at IBM Research - Haifa. This is a list of the current open positions, including internships and and student opportunities.

Meet our interns
Explore

Grants &
fellowship

IBM supports students in their research, encourages and invites the Fellowship winners to do an internship in order to be better informed with the the research done by the company.


Explore

EU funded
projects

IBM Research - Haifa is involved in over 30 EU-funded consortiums and conducts numerous joint projects with academic and industrial partners around the world.


Explore