Pre-training and Fine-tuning Attention Based Encoder Decoder Improves Sea Surface Height Multi-variate Inpainting - Sorbonne Université Access content directly
Conference Papers Year : 2024

Pre-training and Fine-tuning Attention Based Encoder Decoder Improves Sea Surface Height Multi-variate Inpainting

Abstract

The ocean is observed through satellites measuring physical data of various natures. Among them, Sea Surface Height (SSH) and Sea Surface Temperature (SST) are physically linked data involving different remote sensing technologies and therefore different image inverse problems. In this work, we propose to use an Attentionbased Encoder-Decoder to perform the inpainting of the SSH, using the SST as contextual information. We propose to pre-train this neural network on a realistic twin experiment of the observing system and to fine-tune it in an unsupervised manner on real-world observations. We show the interest of this strategy by comparing it to existing methods. Our training methodology achieves state-of-the-art performances, and we report a decrease of 25% in error compared to the most widely used interpolations product.
Fichier principal
Vignette du fichier
visapp2024.pdf (1.49 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04475205 , version 1 (23-02-2024)

Identifiers

  • HAL Id : hal-04475205 , version 1

Cite

Théo Archambault, Arthur Filoche, Anastase Charantonis, Dominique Béréziat. Pre-training and Fine-tuning Attention Based Encoder Decoder Improves Sea Surface Height Multi-variate Inpainting. VISAPP 2024 - 19th International Conference on Computer Vision Theory and Applications, Feb 2024, Roma, Italy. ⟨hal-04475205⟩
81 View
24 Download

Share

Gmail Facebook X LinkedIn More