Neural Architecture Search in operational context: a remote sensing case-study

Anthony Cazasnoves (1) , Pierre-Antoine Ganaye (1) , Kévin Sanchis (1) , Tugdual Ceillier (1)

1 Preligens [Paris]

Deep learning has become in recent years a cornerstone tool fueling key innovations in the industry, such as autonomous driving. To attain good performances, the neural network architecture used for a given application must be chosen with care. These architectures are often handcrafted and therefore prone to human biases and sub-optimal selection. Neural Architecture Search (NAS) is a framework introduced to mitigate such risks by jointly optimizing the network architectures and its weights. Albeit its novelty, it was applied on complex tasks with significant results - e.g. semantic image segmentation. In this technical paper, we aim to evaluate its ability to tackle a challenging operational task: semantic segmentation of objects of interest in satellite imagery. Designing a NAS framework is not trivial and has strong dependencies to hardware constraints. We therefore motivate our NAS approach selection and provide corresponding implementation details. We also present novel ideas to carry out other such use-case studies.

Mots clés: Neural Architecture Search (NAS); High Performance Computing (HPC); Remote Sensing; Semantic segmentation

Domaines: Réseau de neurones [cs.NE]; Intelligence artificielle [cs.AI]; Vision par ordinateur et reconnaissance de formes [cs.CV]

Anthony Cazasnoves, Pierre-Antoine Ganaye, Kévin Sanchis, Tugdual Ceillier. Neural Architecture Search in operational context: a remote sensing case-study. [Research Report] Preligens. 2021. ⟨hal-03343944⟩