EditCLEVR: A Paired-Scene Intervention Benchmark for Compositional Faithfulness of Object-Centric Representations
Quick Answer
EditCLEVR introduces a benchmark for evaluating compositional faithfulness in object-centric representations, focusing on controlled semantic edits.
Quick Take
It features a new metric, Scene-Graph Intervention Accuracy (SGIA), which assesses the correctness of predicted scene changes against intended edits. Baseline evaluations reveal that existing models struggle with semantic faithfulness, particularly in out-of-distribution scenarios.
Key Points
- EditCLEVR features before/after pairs for controlled semantic edits in object-centric learning.
- Introduces Scene-Graph Intervention Accuracy (SGIA) for evaluating semantic change correctness.
- Baseline evaluations indicate performance degradation in CoGenT-OOD-core under ground-truth masks.
- Probes reveal that locality or stability alone can misrepresent semantic faithfulness.
- Code for EditCLEVR is publicly available for further research.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Object-centric learning aims to represent scenes as objects whose properties can be reused in new combinations. Existing evaluations usually score segmentation, single-image factor prediction, or downstream accuracy, but these tests do not directly ask whether a per-object representation behaves correctly under a controlled semantic edit. We introduce EditCLEVR, a paired-scene intervention benchmark in which each example contains a before/after pair of CLEVR-style renders with the same object indices and scene layout, and either exactly one known attribute change on one known object or a no-edit re-render for drift measurement. The protocol includes probe-free diagnostics for representation-change localization and stability, together with probe-decoded semantic faithfulness metrics that test whether the predicted scene change matches the intended intervention across in-distribution and compositional out-of-distribution (OOD) suites, allowing code-space movement and decoded object-attribute correctness to be evaluated separately. We introduce the semantic metric Scene-Graph Intervention Accuracy (SGIA), which requires the full after-scene prediction to be correct and the only predicted before-to-after semantic change to be the intended object-factor edit. We also establish Delta-SGIA as a companion diagnostic that checks the single-site change pattern without requiring the full after-scene graph to be correct. Baseline evaluations on ground-truth-mask backbones, learned-slot models, SAM 2 + frozen-ViT models, and one mask-feature hybrid indicate that CoGenT-OOD-core degradation can persist under ground-truth instance masks, that mask source accounts for part but not all of native performance, and that locality or stability alone can overstate semantic faithfulness. Code is available at this https URL.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2607.22705 [cs.CV] |
| (or arXiv:2607.22705v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2607.22705 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Anuraag Gadehothur Karnam [view email]
[v1]
Sun, 19 Jul 2026 16:11:43 UTC (829 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.