Efficient compression of encoder-decoder models for semantic segmentation using the separation index

We present a novel approach to compressing encoder–decoder architectures, particularly in semantic segmentation tasks, by leveraging the Separation Index (SI)—a metric that quantifies how distinctly a network’s feature maps separate different classes at the pixel level. By identifying and pruning re...

Celý popis

Uložené v:

Podrobná bibliografia
Vydané v:	Scientific reports Ročník 15; číslo 1; s. 24639 - 19
Hlavní autori:	Jamshidi, Movahed, Kalhor, Ahmad, Vahabie, Abdol-Hossein
Médium:	Journal Article
Jazyk:	English
Vydavateľské údaje:	London Nature Publishing Group UK 09.07.2025 Nature Publishing Group Nature Portfolio
Predmet:	639/166/985 639/166/987 Accuracy Compression Encoder-Decoder architectures Humanities and Social Sciences Image processing Methods Model compression multidisciplinary Neural networks Real time Remote sensing Science Science (multidisciplinary) Semantic segmentation Semantics Separation index Tumors Separation index Encoder-Decoder architectures Semantic segmentation Model compression
ISSN:	2045-2322, 2045-2322
On-line prístup:	Získať plný text
Tagy:	Pridať tag Žiadne tagy, Buďte prvý, kto otaguje tento záznam!

Popis
Shrnutí:	We present a novel approach to compressing encoder–decoder architectures, particularly in semantic segmentation tasks, by leveraging the Separation Index (SI)—a metric that quantifies how distinctly a network’s feature maps separate different classes at the pixel level. By identifying and pruning redundant layers and filters, our method preserves the fine-grained spatial details crucial for segmentation while significantly reducing model complexity. We evaluated our approach on five diverse datasets—CamVid (road scenes), KiTS19 (kidney tumor CT scans), the 2018 Data Science Bowl (nuclei segmentation), Aerial Imagery for remote sensing, and MVTec AD (industrial anomaly detection)—across architectures such as U-Net, LinkNet, MobileNet, DeepLabV3, and SegNet. Experimental results show that SI-driven compression reduces parameters and floating-point operations by up to 70% while maintaining or even improving segmentation accuracy, as measured by mean Intersection over Union (IoU). For example, a compressed DeepLabV3 raises the mean IoU from 0.624 to 0.638 on an aerial imagery dataset with a 2.6× reduction in parameters and faster inference. These findings highlight how SI-based pruning balances efficiency and performance, offering a practical solution for resource-constrained semantic segmentation applications.
Bibliografia:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 14 content type line 23
ISSN:	2045-2322 2045-2322
DOI:	10.1038/s41598-025-10348-9