Beyond Unidirectional Bias: Reciprocal Perspective Calibration in Scene Graph Generation

Haifeng Zhao, Wenbo Zhao, Xuemei Luo, Lei-Lei Ma, Dengdi Sun
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:162971-162992, 2026.

Abstract

Scene Graph Generation (SGG) paradigms predominantly model relationships as static, unidirectional mappings ($s \to o$), effectively treating objects as passive recipients of actions. This formulation suffers from an inherent unidirectional bias, violating the physical reality that visual interactions are intrinsically reciprocal. Consequently, existing models often fail to maintain logical self-consistency when the reasoning anchor shifts from the subject to the object. To rectify this cognitive deficiency, we establish the Mutual-Perspective Inverse Relations (MPIR) principle, positing that a robust visual representation must satisfy logical consistency across dual perspectives. Guided by this principle, we propose the Reciprocal Perspective Calibration (RPC) framework, a model-agnostic framework that operationalizes MPIR via a novel Adaptive Inverse-Relation Augmentation (AIRA) strategy. Furthermore, we introduce Hypernym-Guided Prompts (HGP) to bridge the gap between semantic context and computational efficiency in vision-language models, enabling precise modeling of inverse relations. Extensive experiments demonstrate that RPC not only achieves competitive performance on standard benchmarks but also significantly enhances the model’s capability to understand inverse relations, as verified by a new inverse consistency evaluation protocol, demonstrating the cognitive robustness of our method.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-zhao26cb, title = {Beyond Unidirectional Bias: Reciprocal Perspective Calibration in Scene Graph Generation}, author = {Zhao, Haifeng and Zhao, Wenbo and Luo, Xuemei and Ma, Lei-Lei and Sun, Dengdi}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {162971--162992}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/zhao26cb/zhao26cb.pdf}, url = {https://proceedings.mlr.press/v306/zhao26cb.html}, abstract = {Scene Graph Generation (SGG) paradigms predominantly model relationships as static, unidirectional mappings ($s \to o$), effectively treating objects as passive recipients of actions. This formulation suffers from an inherent unidirectional bias, violating the physical reality that visual interactions are intrinsically reciprocal. Consequently, existing models often fail to maintain logical self-consistency when the reasoning anchor shifts from the subject to the object. To rectify this cognitive deficiency, we establish the Mutual-Perspective Inverse Relations (MPIR) principle, positing that a robust visual representation must satisfy logical consistency across dual perspectives. Guided by this principle, we propose the Reciprocal Perspective Calibration (RPC) framework, a model-agnostic framework that operationalizes MPIR via a novel Adaptive Inverse-Relation Augmentation (AIRA) strategy. Furthermore, we introduce Hypernym-Guided Prompts (HGP) to bridge the gap between semantic context and computational efficiency in vision-language models, enabling precise modeling of inverse relations. Extensive experiments demonstrate that RPC not only achieves competitive performance on standard benchmarks but also significantly enhances the model’s capability to understand inverse relations, as verified by a new inverse consistency evaluation protocol, demonstrating the cognitive robustness of our method.} }
Endnote
%0 Conference Paper %T Beyond Unidirectional Bias: Reciprocal Perspective Calibration in Scene Graph Generation %A Haifeng Zhao %A Wenbo Zhao %A Xuemei Luo %A Lei-Lei Ma %A Dengdi Sun %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-zhao26cb %I PMLR %P 162971--162992 %U https://proceedings.mlr.press/v306/zhao26cb.html %V 306 %X Scene Graph Generation (SGG) paradigms predominantly model relationships as static, unidirectional mappings ($s \to o$), effectively treating objects as passive recipients of actions. This formulation suffers from an inherent unidirectional bias, violating the physical reality that visual interactions are intrinsically reciprocal. Consequently, existing models often fail to maintain logical self-consistency when the reasoning anchor shifts from the subject to the object. To rectify this cognitive deficiency, we establish the Mutual-Perspective Inverse Relations (MPIR) principle, positing that a robust visual representation must satisfy logical consistency across dual perspectives. Guided by this principle, we propose the Reciprocal Perspective Calibration (RPC) framework, a model-agnostic framework that operationalizes MPIR via a novel Adaptive Inverse-Relation Augmentation (AIRA) strategy. Furthermore, we introduce Hypernym-Guided Prompts (HGP) to bridge the gap between semantic context and computational efficiency in vision-language models, enabling precise modeling of inverse relations. Extensive experiments demonstrate that RPC not only achieves competitive performance on standard benchmarks but also significantly enhances the model’s capability to understand inverse relations, as verified by a new inverse consistency evaluation protocol, demonstrating the cognitive robustness of our method.
APA
Zhao, H., Zhao, W., Luo, X., Ma, L. & Sun, D.. (2026). Beyond Unidirectional Bias: Reciprocal Perspective Calibration in Scene Graph Generation. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:162971-162992 Available from https://proceedings.mlr.press/v306/zhao26cb.html.

Related Material