{"paper":{"title":"Causal Disentanglement-Inspired Degradation Representation Learning for Full-Reference Image Quality Assessment","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Causal disentanglement separates image content from distortions to enable accurate full-reference quality assessment even without labels.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Fengmao Lv, Jielei Chu, Lin Ma, Tianrui Li, Tian Zhang, Weide Liu, Yuming Fang, Zhen Zhang","submitted_at":"2026-04-23T13:18:13Z","abstract_excerpt":"Existing deep network-based full-reference image quality assessment (FR-IQA) models typically work by performing pairwise comparisons of deep features from the reference and distorted images. In this paper, we approach this problem from a different perspective and propose a novel FR-IQA paradigm based on causal inference and decoupled representation learning. Unlike typical feature comparison-based FR-IQA models, our approach formulates degradation estimation as a causal disentanglement process guided by intervention on latent representations. We first decouple degradation and content represen"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"our method achieves highly competitive performance on standard IQA benchmarks across fully supervised, few-label, and label-free settings. Furthermore, we evaluate the approach on diverse non-standard natural image domains with scarce data... superior cross-domain generalization compared to existing training-free FR-IQA models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"degradation estimation can be formulated as a causal disentanglement process guided by intervention on latent representations, with content invariance between reference and distorted images allowing effective decoupling of degradation and content representations.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Causal disentanglement decouples content and degradation representations via intervention on latents and a content-masking module to predict quality scores from degradation features, achieving strong benchmark performance and cross-domain generalization without labels.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Causal disentanglement separates image content from distortions to enable accurate full-reference quality assessment even without labels.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0b688464874fa94a9e956ddf9e8a0ef673a59bf2a28d8bf78f653519eb478d6c"},"source":{"id":"2604.21654","kind":"arxiv","version":3},"verdict":{"id":"7bca1719-22aa-4e18-93ef-46dc2f040ccf","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-09T22:04:48.069194Z","strongest_claim":"our method achieves highly competitive performance on standard IQA benchmarks across fully supervised, few-label, and label-free settings. Furthermore, we evaluate the approach on diverse non-standard natural image domains with scarce data... superior cross-domain generalization compared to existing training-free FR-IQA models.","one_line_summary":"Causal disentanglement decouples content and degradation representations via intervention on latents and a content-masking module to predict quality scores from degradation features, achieving strong benchmark performance and cross-domain generalization without labels.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"degradation estimation can be formulated as a causal disentanglement process guided by intervention on latent representations, with content invariance between reference and distorted images allowing effective decoupling of degradation and content representations.","pith_extraction_headline":"Causal disentanglement separates image content from distortions to enable accurate full-reference quality assessment even without labels."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.21654/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T12:36:48.392023Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-20T00:46:58.224656Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"9fa6f501d566132954bd3f92959f94425c948604041712e7554e8953c94d671d"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}