Compute-aware evaluation of severity conditioning and lightweight architectures for real-world image deraining

Gunawan, Teddy Surya and Kartiwi, Mira (2026) Compute-aware evaluation of severity conditioning and lightweight architectures for real-world image deraining. IIUM Engineering Journal, 27 (3). pp. 319-339. ISSN 1511-788X E-ISSN 2289-7860

	PDF - Published Version Restricted to Registered users only Download (8MB) | Request a copy
Preview	PDF - Supplemental Material Download (479kB) | Preview
Preview	PDF - Supplemental Material Download (220kB) | Preview

Abstract

Image deraining for embedded perception must balance restoration quality against complete inference cost. Yet density-conditioned methods require annotations unavailable for captured rain, while efficiency studies often omit conditioning-signal cost. This study examines whether label-free severity conditioning justifies its end-to-end cost and whether complexity measures predict realized latency. Under a common L1 protocol, we compare a PReNet baseline with scalar feature-wise linear modulation, spatial severity modulation (SSM), a matched-capacity constant-map control, and two lightweight architectures. Zero-, shuffled-, and constant-prior interventions test SSM’s dependence on prior presence, image correspondence, and spatial structure. Six models were trained for 100 epochs on 10,000 LHP-Rain patches using three seeds and evaluated on the 1,000-image test split. SSM achieved 31.775±0.109 dB, compared with 31.741±0.095 dB for Base and 31.754±0.102 dB under zero-prior replacement. A seed-level equivalence analysis based on paired image differences supported practical equivalence between SSM and Base (95% CI: ?0.026 to 0.094 dB; margin fixed before primary analysis: ±0.1 dB), consistent with the half-scale experiment. The two-stage model reduced computation from 265.0 to 44.6 GFLOPs and model latency from 20.43 to 4.50 ms, but the prior generation increased total latency to 24.74 ms. CBAM-light required 14.1 GFLOPs yet recorded 33.68 ms. The label-free statistic achieved a cross-scene AUC of 0.772 on RealRain-1k, indicating moderate agreement with density labels. Thus, severity-conditioned and lightweight deraining systems should be assessed through verified prior use and measured end-to-end latency rather than accuracy, parameters, or GFLOPs alone.

Item Type:	Article (Journal)
Uncontrolled Keywords:	Image Deraining, Severity Conditioning, Lightweight Architecture, End-to-End Latency, Real-World Rain
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800 Electronics. Computer engineering. Computer hardware. Photoelectronic devices > TK7885 Computer engineering
Kulliyyahs/Centres/Divisions/Institutes (Can select more than one option. Press CONTROL button):	Kulliyyah of Engineering > Department of Electrical and Computer Engineering Kulliyyah of Information and Communication Technology > Department of Information System Kulliyyah of Information and Communication Technology > Department of Information System
Depositing User:	Prof. Dr. Teddy Surya Gunawan
Date Deposited:	28 Sep 2026 14:40
Last Update:	28 Sep 2026 14:40
Queue Number:	2026-09-Q5298
URI:	http://irep.iium.edu.my/id/eprint/131545
Indexed In:	WOS and SCOPUS, ERA, Google Scholar, MyCite

Actions (login required)

View Item