RLDF (Region-Level Deepfake) is the first dataset designed to evaluate deepfake detection on partial facial manipulations. It contains over 2 million partial face‑swap images generated with eight face‑swapping methods across seven facial regions, including both discrete (eye, nose, mouth, eyebrow) and composite regions. The authors benchmark 12 state‑of‑the‑art detectors on five cross‑dataset protocols, revealing a 10.5 % drop in AUC and 10 % drop in accuracy for region‑level detection, and propose a mask‑guided cross‑attention model as a baseline.
By Shukesh Reddy, Abhijit Das
The study examines whether adding illumination inconsistencies to self-blended images—creating Temporal Self-Blended Images (T‑SBI)—enhances face‑swap detection. Across five training regimes and multiple datasets, no significant improvement in detection accuracy was observed; AUC differences stayed within seed variability. However, T‑SBI altered prediction score distributions, shifting optimal thresholds, and did improve robustness to heavy JPEG compression on the DFDC dataset, suggesting a reliance on low‑frequency cues.
By Danil Davydov, Bader Rasheed, Dmitriy Vatolin
arXiv:2609.01511v1 Announce Type: new
Abstract: Face forgery detectors often achieve strong results on controlled benchmarks, but their reliability under realistic image degradations remains limited....
By Lucas Cunha, Lucas Sotomaior, Lucas Gasperin, Beatriz Caldas, Eduardo Pianovski, Rayson Laroca
arXiv:2609.12668v1 Announce Type: new
Abstract: Recent deepfake detection studies increasingly suggest remote photoplethysmography (rPPG) signals as an authenticity cue. However, existing benchmarks...
By Chenxi Yang, Yassine Ouzar, Larbi Boubchir
RA‑SOD is a new RGB‑Thermal salient object detection framework that explicitly models the reliability of each modality. It introduces a reliability‑conditioned representation, an uncertainty‑guided dual‑stream refinement, and a pixel‑wise modality competition mechanism to adaptively compensate degraded features and suppress unreliable evidence. Experiments on four benchmarks show that RA‑SOD achieves state‑of‑the‑art performance and remains robust under severe modality degradation.
By Hongbo Gao, Zhengyu Li, Xueru Nie, Dihao Zhu, Lijun Zhao, Yunke Wang, Chang Xu
The paper investigates how knowledge distillation from event cameras to RGB images can alter the inductive biases of convolutional neural networks. By transferring learning from the event domain, the authors find that models gain color invariance, a shape bias, and improved robustness to high‑frequency noise, largely due to reduced reliance on texture and increased emphasis on edge‑based object shape. These changes are evidenced by early‑layer processing differences and a spectral trade‑off between robustness to missing high‑frequency content and vulnerability to its contamination or geometric disruption.
By Soshun Kihara, Shunsuke Yasuki, Masato Taki