SHAMISA: SHAped Modeling of Implicit Structural Associations for Self-supervised No-Reference Image Quality Assessment

Mahdi Naseri; Zhou Wang

cs.CVMar 14, 2026

SHAMISA: SHAped Modeling of Implicit Structural Associations for Self-supervised No-Reference Image Quality Assessment

Authors: Mahdi Naseri, Zhou Wang

Organizations: Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada

Abstract

No-Reference Image Quality Assessment (NR-IQA) aims to estimate perceptual quality without access to a reference image of pristine quality. Learning an NR-IQA model faces a fundamental bottleneck: its need for a large number of costly human perceptual labels. We propose SHAMISA, a non-contrastive self-supervised framework that learns from unlabeled distorted images by leveraging explicitly structured relational supervision. Unlike prior methods that impose rigid, binary similarity constraints, SHAMISA introduces implicit structural associations, defined as soft, controllable relations that are both distortion-aware and content-sensitive, inferred from synthetic metadata and intrinsic feature structure. A key innovation is our compositional distortion engine, which generates an uncountable family of degradations from continuous parameter spaces, grouped so that only one distortion factor varies at a time. This enables fine-grained control over representational similarity during training: images with shared distortion patterns are pulled together in the embedding space, while severity variations produce structured, predictable shifts. We integrate these insights via dual-source relation graphs that encode both known degradation profiles and emergent structural affinities to guide the learning process throughout training. A convolutional encoder is trained under this supervision and then frozen for inference, with quality prediction performed by a linear regressor on its features. Extensive experiments on synthetic, authentic, and cross-dataset NR-IQA benchmarks demonstrate that SHAMISA achieves strong overall performance with improved cross-dataset generalization and robustness, all without human quality annotations or contrastive losses.

Figures & tables

Fig. 1: Overview of the proposed SHAMISA framework. (1) Self-supervised pre-training: pristine images are transformed by a compositional distortion engine to form a mini-batch $\mathbf{X}$ and distortion metadata; the encoder $f_{\theta}$ and projector $g_{\psi}$ produce representations $\mathbf{H}$ and embeddings $\mathbf{Z}$ used to construct dual-source relation graphs (metadata-driven and structurally intrinsic), which are aggregated by a multi-source aggregator $\Phi$ into $\mathbf{G}$ and optimize a graph-weighted SSL objective. (2) Downstream NR-IQA: we freeze $f_{\theta}$ and train a lightweight regressor on top of $\mathbf{H}$ to predict quality scores.

		Synthetic distortions								Authentic distortions
Method	Type	LIVE		CSIQ		TID2013		KADID-10K		FLIVE		SPAQ		Average
		SRCC	PLCC	SRCC	PLCC	SRCC	PLCC	SRCC	PLCC	SRCC	PLCC	SRCC	PLCC	SRCC	PLCC
BRISQUE [ 20 ]	Handcrafted	0.939	0.935	0.746	0.829	0.604	0.694	0.528	0.567	0.288	0.373	0.809	0.817	0.652	0.703
NIQE [ 21 ]	Handcrafted	0.907	0.901	0.627	0.712	0.315	0.393	0.374	0.428	0.211	0.288	0.700	0.709	0.522	0.572
CORNIA [ 24 ]	Codebook	0.947	0.950	0.678	0.776	0.678	0.768	0.516	0.558	–	–	0.709	0.725	–	–
HOSA [ 25 ]	Codebook	0.946	0.950	0.741	0.823	0.735	0.815	0.618	0.653	–	–	0.846	0.852	–	–

TABLE I: Performance comparison on synthetic and in-the-wild NR-IQA datasets. We report SRCC and PLCC. Best and second-best per column are bold and underlined . Prior numbers are taken from their papers or public code where available [ 17 , 11 , 18 ] .

Training	Testing	HyperIQA	Su et al.	CONTRIQUE	Re-IQA	ARNIQA	SHAMISA
LIVE	CSIQ	0.744	0.777	0.803	0.795	0.904	0.909
LIVE	TID2013	0.541	0.561	0.640	0.588	0.697	0.700
LIVE	KADID-10K	0.492	0.506	0.699	0.557	0.764	0.694
CSIQ	LIVE	0.926	0.930	0.912	0.919	0.921	0.939
CSIQ	TID2013	0.541	0.550	0.570	0.575	0.721	0.729
CSIQ	KADID-10K	0.509	0.515	0.696	0.521	0.735	0.711

TABLE II: Cross-dataset SRCC on synthetic NR-IQA benchmarks. Best and second-best per row are bold and underlined . Baseline values are taken from prior reports in ARNIQA and Re-IQA [ 18 , 11 ] .

Panel of Figure 4 — Fig. 2: gMAD on Waterloo Exploration (distorted pool): panels (a,b) use SHAMISA as the defender and panels (c,d) use ARNIQA as the defender. In each panel, the top and bottom images are the attacker-selected pair from the indicated defender-quality bin, with low-bin cases shown in (a,c) and high-bin cases shown in (b,d). Supplementary Table VII reports the corresponding top-1 attacker gaps.

Fig. 3: t-SNE visualization of image-level encoder representations on KADID-10K. We extract encoder representations $\mathbf{H}$ (pre-projector) and average across crops to obtain one representation per image, yielding 10,125 image-level points. Points are colored by the coarse distortion family using a shared fixed palette. Compared to ARNIQA, SHAMISA exhibits clearer coarse-family structure for several dominant degradation families, while both models show overlap among visually related families such as brightness and color manipulations. This visualization is a qualitative diagnostic and does not affect the quantitative evaluation protocol.

Panel of Figure 6 — Fig. 4: Manifold visualization with UMAP [ 51 ] using encoder representations $\mathbf{H}$ extracted from 1,000 pristine KADIS images degraded with Gaussian blur and white noise. For each pristine image we generate 5 blur-only samples, 5 noise-only samples, and all $5\times 5$ blur $\rightarrow$ noise compositions (35 variants per image, 35k total points). Point color is the weighted average of blur (red) and noise (yellow) according to their relative intensities; marker opacity increases with total severity.

(A) Aggregation & regularization
Variant	SRCC g	PLCC g	SRCC a	PLCC a	SRCC s	PLCC s
A0: SHAMISA	0.8862	0.9042	0.7620	0.8040	0.9483	0.9543
A1: Binary ${\bm{G}}$	0.8767	0.8924	0.7475	0.7821	0.9374	0.9457
A3: No $\Phi$ (fixed linear mixture)	0.8759	0.8951	0.7479	0.7896	0.9361	0.9457
(B) Graph sources
B1: No ${\bm{G}}^{o}$ (OT graph)	0.8600	0.8787	0.7661	0.8069	0.8995	0.9092

TABLE III: Main ablations . Median over 10 splits per dataset; then averaged. SRCC s /PLCC s average over synthetic

\{\mathrm{LIVE},\mathrm{CSIQ},\mathrm{TID2013},\mathrm{KADID\mbox{-}10K}\}

, SRCC a /PLCC a over authentic

\{\mathrm{FLIVE},\mathrm{SPAQ}\}

, and SRCC g /PLCC g denote the average over the six per-dataset median SRCC/PLCC values. Full ablation catalog is reported in Supplementary Table VIII.

Fig. 5: SRCC vs. training steps during SSL pre-training (mean with

\pm

std error bars over 10 splits) across representative NR-IQA benchmarks. Curves typically rise early and then saturate, while the best checkpoint can vary by dataset since the SSL objective is not identical to downstream SRCC.

Fig. 6: One-at-a-time sweeps around the SHAMISA reference A0, reporting SRCC across six NR-IQA benchmarks in a single merged view. Columns correspond to

(d_{h},d_{z})

(B,\xi)

(\alpha,\beta)

(\gamma,\eta)

, and

(k_{n},K)

, with a shared legend across all panels.

Appendix figures & tables11 assets

Supplementary material from the paper’s appendix.

Appendix

Group	Hyperparameters
Architecture	ResNet-50 encoder; $d_{h}=2048$ ; projector output $d_{z}=256$ .
Optimization	SGD with $\mathrm{lr}=1.5\times 10^{-3}$ , momentum $0.9$ , weight decay $10^{-4}$ , and cosine annealing warm restarts.
Loss	VICReg-style weights $(\alpha,\beta,\gamma)=(11.98,57.21,88.37)$ ; OT alignment weight $\eta=0.4906$ ; graph regularizer weight $\xi=0.0342$ .
Graphs	kNN graph: $k_{n}=31$ ; metadata graph sparsifier: $K_{d}=4096$ ; ref-ref edge weight $w_{rr}=0.5766$ ; OT graph: $K=32$ , $\tau_{c}=0.1$ ; OT global sparsifier: $K_{g}=8N$ .
Distortion engine	Per update: $B=2$ tiny-batches, $R=3$ references, $C=4$ composition groups, $L=5$ severity levels; crop size $224$ ; max distortions per sample $M_{d}=7$ .

Appendix

TABLE IV: Essential final hyperparameters used for SHAMISA pre-training. A single configuration is shared across all datasets (no dataset-specific tuning during SSL).

Dataset	Type	Pristine refs	Distortion types	Images	Label type	Split note
LIVE [ 41 ]	Synthetic	29	5	779	DMOS	Reference-disjoint
CSIQ [ 42 ]	Synthetic	30	6	866	DMOS	Reference-disjoint
TID2013 [ 43 ]	Synthetic	25	24	3000	MOS	Reference-disjoint
KADID-10K [ 9 ]	Synthetic	81	25	10,125	MOS	Reference-disjoint
FLIVE [ 29 ]	In-the-wild	–	–	39,808	MOS + patch scores	Official split
SPAQ [ 44 ]	In-the-wild	–	–	11,125	MOS	Random 70/10/20

Appendix

TABLE V: Dataset statistics for the NR-IQA benchmarks. Synthetic datasets are reported by reference count and distortion inventory, whereas authentic datasets are summarized by image count and label type.

Method	Type	LIVE		CSIQ		TID2013		KADID-10K
		SRCC	PLCC	SRCC	PLCC	SRCC	PLCC	SRCC	PLCC
PSNR	Traditional	0.881	0.868	0.820	0.824	0.643	0.675	0.677	0.680
SSIM [ 52 ]	Traditional	0.921	0.911	0.854	0.835	0.642	0.698	0.641	0.633
FSIM [ 53 ]	Traditional	0.964	0.954	0.934	0.919	0.852	0.875	0.854	0.850
VSI [ 54 ]	Traditional	0.951	0.940	0.944	0.929	0.902	0.903	0.880	0.878
PieAPP [ 55 ]	Deep learning	0.915	0.905	0.900	0.881	0.877	0.850	0.869	0.869

Appendix

TABLE VI: Full-reference IQA comparison on synthetic distortion datasets. Reported rows are taken from prior work. SHAMISA-FR uses the same fixed-encoder evaluation protocol as our NR setting, and each value is the median over 10 random splits. Best and second-best values in each column are shown in bold and underlined.

Fig. 7: t-SNE colored by distortion severity on KADID-10K. Colors indicate severity levels 1 to 5 for ARNIQA (left) and SHAMISA (right).

Fig. 8: Subtype-colored t-SNE views for SHAMISA, including six KADID-10K distortion families.

Defender, Quality Level	Bin size	Top-1 attacker gap
ARNIQA, high	9488	8.2717
ARNIQA, low	9488	8.6782
SHAMISA, high	9488	4.1860
SHAMISA, low	9488	3.1356

Appendix

TABLE VII: gMAD summary on Waterloo Exploration (distorted pool) between SHAMISA and ARNIQA. Attacker gap denotes the attacker-side score separation of the selected pair within the defender-defined bin; lower is better.

(A) Graph weighting & regularization
Variant	SRCC g	PLCC g	SRCC a	PLCC a	SRCC s	PLCC s
A0: SHAMISA	0.8862	0.9042	0.7620	0.8040	0.9483	0.9543
A1: Binary ${\bm{G}}$	0.8767	0.8924	0.7475	0.7821	0.9374	0.9457
A2: No $\mathcal{R}_{\mathrm{graph}}$	0.8813	0.8981	0.7471	0.7908	0.9450	0.9497
A3: No $\Phi$ (fixed linear mixture)	0.8759	0.8951	0.7479	0.7896	0.9361	0.9457
A4: No stop-grad through $\Phi$	0.8797	0.8980	0.7521	0.7950	0.9395	0.9469

Appendix

TABLE VIII: Full ablation catalog. For each dataset we run 10 random splits and report the median test SRCC/PLCC across splits, then average within groups: synthetic

\{\mathrm{LIVE},\mathrm{CSIQ},\mathrm{TID2013},\mathrm{KADID\mbox{-}10K}\}

, authentic

\{\mathrm{FLIVE},\mathrm{SPAQ}\}

, and SRCC g /PLCC g denote the average over the six per-dataset median SRCC/PLCC values.

Panel of Figure 17 — Fig. 9: Feature decorrelation proxy ( corr ) over SSL pre-training. Both ${\bm{H}}$ and ${\bm{Z}}$ show a rapid early drop followed by stabilization, consistent with progressively reduced redundancy across feature dimensions.

Fig. 10: Effective dimensionality proxy ( rank ) for

{\bm{H}}

over SSL pre-training. The encoder representation uses a substantial number of active dimensions throughout training and then stabilizes after the early phase.

Panel of Figure 19 — Fig. 11: Invariance proxy ( inv ) over SSL pre-training. Curves generally decrease and then stabilize, consistent with paired views becoming closer in representation space as training proceeds.

HP	LIVE	CSIQ	TID2013	KADID-10K	FLIVE	SPAQ	Mean
$d_{h}$	0.012	0.003	0.003	0.006	0.004	0.007	0.006
$d_{z}$	0.006	0.002	0.002	0.000	0.003	0.007	0.003
$B$	0.002	0.001	0.003	0.005	0.001	0.001	0.002
$k_{n}$	0.007	0.009	0.006	0.016	0.007	0.001	0.008
$K$	0.019	0.035	0.061	0.030	0.007	0.005	0.026
$\gamma$	0.017	0.045	0.089	0.070	0.033	0.001	0.042

Appendix

TABLE IX: Sensitivity summary for SRCC. For each hyperparameter,

\Delta

is the largest minus smallest SRCC across the sweep. Larger

\Delta

indicates higher sensitivity. The most sensitive and second most sensitive hyperparameters per dataset are shown in bold and underlined .

Explore similar work

Preference-Guided Debiasing for No-Reference Enhancement Image Quality Assessment
Mar 20, 2026Shiqi Gao, Kang Fu, Zitong Xu +4Image Quality Assessment Debiased ML
Institute of Image Communication and Network Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
Pixel Perfect: Relational Image Quality Assessment with Spatially-Aware Distortions
May 4, 2026Fadeel Sher Khan, Long N. Le, Abhinau K. Venkataramanan +2Image Quality Assessment
The University of Texas at Austin Austin, TX, USA · Samsung Research America Plano, TX, USA
Spatially Localized Image Degradation Embeddings for Image Quality Assessment
Jun 28, 2026Krishna Srikar Durbha, Hassene Tmar, Ping-Hao Wu +2Image Quality Assessment Self-Supervised Visual Representation Learning
The University of Texas at Austin · Meta Platforms, Inc. · University of Colorado Boulder

Mar 20, 2026J/K move · Enter open · S save

Preference-Guided Debiasing for No-Reference Enhancement Image Quality Assessment

Shiqi Gao, Kang Fu, Zitong Xu, Huiyu Duan+3

Institute of Image Communication and Network Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

Image Quality Assessment Debiased ML Image Enhancement

Open paper AskPDF arXiv

Current no-reference image quality assessment (NR-IQA) models for enhanced images often struggle to generalize, as they tend to overfit to the distinct patterns of specific enhancement algorithms rather than evaluating genuine perceptual quality. To address this issue, we propose a preference-guided debiasing framework for no-reference enhancement image quality assessment (EIQA). Specifically, we first learn a continuous enhancement-preference embedding space using supervised contrastive learning, where images generated by similar enhancement styles are encouraged to have closer representations. Based on this, we further estimate the enhancement-induced nuisance component contained in the raw quality representation and remove it before quality regression. In this way, the model is guided to focus on algorithm-invariant perceptual quality cues instead of enhancement-specific visual fingerprints. To facilitate stable optimization, we adopt a two-stage training strategy that first learns the enhancement-preference space and then performs debiased quality prediction. Extensive experiments on public EIQA benchmarks demonstrate that the proposed method effectively mitigates algorithm-induced representation bias and achieves superior robustness and cross-algorithm generalization compared with existing approaches.

Abstract

Explore similar work

Preference-Guided Debiasing for No-Reference Enhancement Image Quality Assessment

Pixel Perfect: Relational Image Quality Assessment with Spatially-Aware Distortions

Spatially Localized Image Degradation Embeddings for Image Quality Assessment