Salvato in:
Dettagli Bibliografici
Autori principali: Dekel, Shay, Keller, Yosi, Cadik, Martin
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:https://arxiv.org/abs/2303.02615
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913258418470912
author Dekel, Shay
Keller, Yosi
Cadik, Martin
author_facet Dekel, Shay
Keller, Yosi
Cadik, Martin
contents The estimation of large and extreme image rotation plays a key role in multiple computer vision domains, where the rotated images are related by a limited or a non-overlapping field of view. Contemporary approaches apply convolutional neural networks to compute a 4D correlation volume to estimate the relative rotation between image pairs. In this work, we propose a cross-attention-based approach that utilizes CNN feature maps and a Transformer-Encoder, to compute the cross-attention between the activation maps of the image pairs, which is shown to be an improved equivalent of the 4D correlation volume, used in previous works. In the suggested approach, higher attention scores are associated with image regions that encode visual cues of rotation. Our approach is end-to-end trainable and optimizes a simple regression loss. It is experimentally shown to outperform contemporary state-of-the-art schemes when applied to commonly used image rotation datasets and benchmarks, and establishes a new state-of-the-art accuracy on these datasets. We make our code publicly available.
format Preprint
id arxiv_https___arxiv_org_abs_2303_02615
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Estimating Extreme 3D Image Rotation with Transformer Cross-Attention
Dekel, Shay
Keller, Yosi
Cadik, Martin
Computer Vision and Pattern Recognition
The estimation of large and extreme image rotation plays a key role in multiple computer vision domains, where the rotated images are related by a limited or a non-overlapping field of view. Contemporary approaches apply convolutional neural networks to compute a 4D correlation volume to estimate the relative rotation between image pairs. In this work, we propose a cross-attention-based approach that utilizes CNN feature maps and a Transformer-Encoder, to compute the cross-attention between the activation maps of the image pairs, which is shown to be an improved equivalent of the 4D correlation volume, used in previous works. In the suggested approach, higher attention scores are associated with image regions that encode visual cues of rotation. Our approach is end-to-end trainable and optimizes a simple regression loss. It is experimentally shown to outperform contemporary state-of-the-art schemes when applied to commonly used image rotation datasets and benchmarks, and establishes a new state-of-the-art accuracy on these datasets. We make our code publicly available.
title Estimating Extreme 3D Image Rotation with Transformer Cross-Attention
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2303.02615