End-to-End Audio Strikes Back: Boosting Augmentations Towards An Efficient Audio Classification Network

1 code implementation • 25 Apr 2022 • Avi Gazneli, Gadi Zimerman, Tal Ridnik, Gilad Sharir, Asaf Noy

While efficient architectures and a plethora of augmentations for end-to-end image classification tasks have been suggested and heavily investigated, state-of-the-art techniques for audio classifications still rely on numerous representations of the audio signal together with large architectures, fine-tuned from large datasets.

Environmental Sound Classification Keyword Spotting +2

ML-Decoder: Scalable and Versatile Classification Head

1 code implementation • 25 Nov 2021 • Tal Ridnik, Gilad Sharir, Avi Ben-Cohen, Emanuel Ben-Baruch, Asaf Noy

In this paper, we introduce ML-Decoder, a new attention-based classification head.

Classification Decoder +3

PETA: Photo Albums Event Recognition using Transformers Attention

1 code implementation • 26 Sep 2021 • Tamar Glaser, Emanuel Ben-Baruch, Gilad Sharir, Nadav Zamir, Asaf Noy, Lihi Zelnik-Manor

We address this gap with a tailor-made solution, combining the power of CNNs for image representation and transformers for album representation to perform global reasoning on image collection, offering a practical and efficient solution for photo albums event recognition.

An Image is Worth 16x16 Words, What is a Video Worth?

2 code implementations • 25 Mar 2021 • Gilad Sharir, Asaf Noy, Lihi Zelnik-Manor

Methods that reach State of the Art (SotA) accuracy, usually make use of 3D convolution layers as a way to abstract the temporal information from video frames.

Action Classification Action Recognition

TResNet: High Performance GPU-Dedicated Architecture

3 code implementations • 30 Mar 2020 • Tal Ridnik, Hussam Lawen, Asaf Noy, Emanuel Ben Baruch, Gilad Sharir, Itamar Friedman

In this work, we introduce a series of architecture modifications that aim to boost neural networks' accuracy, while retaining their GPU training and inference efficiency.

Fine-Grained Image Classification General Classification +5

Graph Embedded Pose Clustering for Anomaly Detection

1 code implementation • CVPR 2020 • Amir Markovitz, Gilad Sharir, Itamar Friedman, Lihi Zelnik-Manor, Shai Avidan

We propose a new method for anomaly detection of human actions.

Anomaly Detection Clustering +1

Video Object Segmentation using Tracked Object Proposals

no code implementations • 20 Jul 2017 • Gilad Sharir, Eddie Smolyansky, Itamar Friedman

We present an approach to semi-supervised video object segmentation, in the context of the DAVIS 2017 challenge.

Object object-detection +6