Publications by Amir Roshan Zamir

How Far Can a 1-Pixel Camera Go? Solving Vision Tasks Using Photoreceptors and Computationally Designed Visual Morphology

A de facto standard approach in solving computer vision tasks is to use a common high-resolution camera and choose its placement on an agent based on human intuition. On the other hand, extremely simple and well-designed visual sensors found throughout nat ...

Springer Science and Business Media Deutschland GmbH2025

ViPer: Visual Personalization of Generative Models via Individual Preference Learning

Amir Roshan Zamir, Roman Christian Bachmann, Shuqing Teresa Yeo

Different users find different images generated for the same prompt desirable. This gives rise to personalized image generation which involves creating images aligned with an individual’s visual preference. Current generative models are, however, tuned to ...

Springer Science and Business Media Deutschland GmbH2025

BRAVE: Broadening the Visual Encoding of Vision-Language Models

Amir Roshan Zamir

Vision-language models (VLMs) are typically composed of a vision encoder, e.g. CLIP, and a language model (LM) that interprets the encoded features to solve downstream tasks. Despite remarkable progress, VLMs are subject to several shortcomings due to the ...

Springer Science and Business Media Deutschland GmbH2025

4M: Massively Multimodal Masked Modeling

Amir Roshan Zamir, Oguzhan Fatih Kar, Roman Christian Bachmann, Shuqing Teresa Yeo, David Mizrahi

Current machine learning models for vision are often highly specialized and limited to a single modality and task. In contrast, recent large language models exhibit a wide range of capabilities, hinting at a possibility for similarly versatile models in co ...

Neural Information Processing Systems (Nips)2023

Modality-invariant Visual Odometry for Embodied Vision

Amir Roshan Zamir, Roman Christian Bachmann, Marius Reinhard Memmel

Effectively localizing an agent in a realistic, noisy setting is crucial for many embodied vision tasks. Visual Odometry (VO) is a practical substitute for unreliable GPS and compass sensors, especially in indoor environments. While SLAM-based methods show ...

Ieee Computer Soc2023

Rapid Network Adaptation: Learning to Adapt Neural Networks Using Test-Time Feedback

Amir Roshan Zamir, Oguzhan Fatih Kar, Shuqing Teresa Yeo, Zahra Sodagar

We propose a method for adapting neural networks to distribution shifts at test-time. In contrast to training-time robustness mechanisms that attempt to anticipate and counter the shift, we create a closed-loop system and make use of test-time feedback sig ...

Ieee Computer Soc2023

MultiMAE: Multi-modal Multi-task Masked Autoencoders

Amir Roshan Zamir, Roman Christian Bachmann, Andrei Atanov, David Mizrahi

We propose a pre-training strategy called Multi-modal Multi-task Masked Autoencoders (MultiMAE). It differs from standard Masked Autoencoding in two key aspects: I) it can optionally accept additional modalities of information in the input besides the RGB ...

SPRINGER INTERNATIONAL PUBLISHING AG2022

3D Common Corruptions and Data Augmentation

Amir Roshan Zamir, Oguzhan Fatih Kar, Andrei Atanov, Shuqing Teresa Yeo

We introduce a set of image transformations that can be used as corruptions to evaluate the robustness of models as well as data augmentation mechanisms for training neural networks. The primary distinction of the proposed transformations is that, unlike e ...

IEEE COMPUTER SOC2022

CLIPasso: Semantically-Aware Object Sketching

Amir Roshan Zamir, Roman Christian Bachmann, Ehsan Pajouheshgar

The Abstraction is at the heart of sketching due to the simple and minimal nature of line drawings. Abstraction entails identifying the essential visual properties of an object or scene, which requires semantic understanding and prior knowledge of high-lev ...

ASSOC COMPUTING MACHINERY2022

Robustness via Cross-Domain Ensembles

Amir Roshan Zamir, Oguzhan Fatih Kar, Shuqing Teresa Yeo

We present a method for making neural network predictions robust to shifts from the training data distribution. The proposed method is based on making predictions via a diverse set of cues (called 'middle domains') and ensembling them into one strong predi ...

IEEE2021