- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
ViGuardQ addresses post-training quantization of video deepfake detectors. Dual-Evidence-Constrained Rotation identifies spatial and temporal decision-sensitive feature directions and learns rotations and quantization scales while preserving the selected subspaces. Evidence-guided Consistent Response calibration subsequently matches the full-precision detector's responses to temporally aligned and opposing feature perturbations. The original detector weights remain frozen, with calibration performed on 256 videos. Experiments cover eleven detectors and six benchmarks, reporting improved AUC over quantization baselines and an Android deployment demonstration.
Critical strengths
-
Preserving decision-sensitive directions and perturbation responses is a reasonable approach to limiting quantization damage to subtle forensic signals. The method operates on existing detectors without retraining their original weights, making the calibration setting practically relevant.
-
The reported improvements extend across multiple architectures and datasets. The component ablations also provide useful evidence that DEC and ECR contribute to the reported gains.
Critical weaknesses
-
The additional contribution beyond existing task-specific quantization is not sufficiently established. Preserving subtle forgery features during low-bit quantization is already central to Tao et al., "Unlocking the Potential of Lightweight Quantized Models for Deepfake Detection" (IJCAI 2025), which this submission cites. ViGuardQ's PTQ setting and temporal constraints provide plausible distinctions, but the evidence for the latter remains limited. The evaluation lists DeiT, ViT, Swin, and existing detectors without clearly specifying their temporal architectures, frame aggregation, or feature correspondence construction. Consequently, it is unclear whether the method preserves genuinely temporal detection behavior or improves calibration of predominantly framewise models. Appendix Table 6 shows benefits from adding TOP to TAP, but does not isolate temporal structure from additional perturbation coverage or calibration computation. A comparison against randomly paired or correspondence-shuffled perturbations under the same calibration budget would be substantially more informative.
-
The reported accuracy-efficiency advantage is difficult to assess under the specified experimental conditions. The results do not clearly identify the weight and activation bit-widths, excluded full-precision layers, or whether quantization granularity is matched across methods. These choices directly affect both accuracy and resource use, so the large baseline gaps cannot yet be attributed confidently to the proposed design. Similarly, the Android figures of 90 ms per frame and 92 mW lack sufficient device, runtime, input, and measurement details, together with matched full-precision and quantized baseline measurements. DEC retains runtime rotations, but their contribution to the reported latency is not quantified. These omissions leave the central deployment claim insufficiently supported by the reported storage reductions alone.
Questions for authors
-
Which evaluated detectors explicitly model temporal interactions, and how are matched feature locations obtained?
-
Does the temporal benefit persist against random-pairing controls with matched perturbation counts and calibration budgets?
-
What exact weight/activation precisions, quantization groups, and full-precision exceptions are used for each method? Are the comparisons matched on these settings?
6 comments
Response to Reviewer APAa's Weaknesses 1 and Question 1 & 2 (Continued)
W1.2 Clarification of Temporal Infomation of Evaluated Detectors.
First, most advanced video deepfake detectors use image-based ViT backbones, encoding temporal information through input representations without modifying the underlying Transformer architecture [1–3]. Our evaluated detectors follow the same approach, incorporating temporal information through temporal encoding fusion while retaining the image-based ViT architectures. This motivates ViGuardQ to preserve discriminative spatial and temporal forgery evidence during quantization.
Second, to further demonstrate the generalizability of ViGuardQ, we evaluate two video-based transformers, ViViT[4] and FakeSTormer[5]: Both extend image-based vision transformer architectures with explicit temporal modeling mechanisms to capture cross-frame dependencies. Our results demonstrate that ViGuardQ can effectively quantize temporally enhanced detectors while maintaining competitive detection performance (Table 1 here).
Table 1. Evaluation on Video-based ViT Detectors (AUC %) on DFD dataset.
| Dataset | Detector | Metric | FP (GT) | BRECQ | AdaLog | FIMA-Q | PTQ4VM | ViGuardQ |
|---|---|---|---|---|---|---|---|---|
| DFD | ViViT | Storage Cost (MB) ↓ | 78.6 | 46.2 | 42.5 | 41.1 | 43.0 | |
| Compression Rate (%) ↓ | - | 22.1 | 13.0 | 12.0 | 11.6 | 12.1 | ||
| Detection AUC (%) ↑ | 56.3 | 60.1 | 62.0 | 67.6 | 76.3 | |||
| FakeSTormer | Storage Cost (MB) ↓ | 113.6 | 61.9 | 60.0 | 58.2 | 60.4 | ||
| Compression Rate (%) ↓ | - | 22.2 | 12.1 | 11.7 | 11.4 | 11.8 | ||
| Detection AUC (%) ↑ | 59.2 | 72.3 | 73.1 | 79.4 | 89.6 |
Last, Our method explicitly uses temporal information during quantization process.
Theoretically, ViGuardQ operates on the input and output representations of detector blocks, enabling it to handle temporal information whether encoded at the feature level or explicitly modeled by the detector architecture. Specifically, DEC combines within-frame decision gradients and cross-frame gradient differences at corresponding facial locations (Eq. 6) to identify channels sensitive to spatial and temporal evidence. ECR uses temporally aligned and opposed feature perturbations to preserve the FP detector's signed changes in video-level decision margins. Together, these objectives enable ViGuardQ to preserve temporally relevant forensic evidence across different detector architectures.
Experimentally, we compare adjacent and nonadjacent frame pairings within the same facial tracks during DEC and ECR calibration to evaluate the contribution of temporal adjacency (Table 2 here). According to the results, using adjacent rather than nonadjacent pairs in both modules improves video-level AUC performance. Moreover, changing only ECR from adjacent to nonadjacent pairing reduces AUC value. These results indicate that temporally local frame correspondences are beneficial for preserving the FP detector's decision behavior during quantization.
Table 2: Ablation Study of Temporal Pairing in DEC and ECR (AUC %).
| DFD | DFDC | |||||||
|---|---|---|---|---|---|---|---|---|
| Setting | DEC | ECR | DeiT-B | ViT-B | Swin-B | DeiT-B | ViT-B | Swin-B |
| FP | - | - | 88.8 | 93.7 | 93.2 | 75.6 | 77.8 | 77.4 |
| A. Base | - | - | 52.2 | 54.9 | 51.9 | 42.3 | 45.8 | 43.4 |
| B1. DEC only | ✓ Adjacent | - | 73.6 | 71.9 | 75.6 | 60.9 | 62.7 | 64.1 |
| B2. DEC only | ✓ Nonadjacent | - | 57.5 | 57.0 | 53.2 | 47.6 | 49.2 | 50.0 |
| C1. Change both pairings | ✓ Nonadjacent | ✓ Nonadjacent | 68.2 | 65.4 | 70.3 | 52.1 | 50.4 | 56.7 |
| C2. Change ECR pairing | ✓ Adjacent | ✓ Nonadjacent | 75.5 | 70.0 | 77.7 | 60.0 | 65.3 | 61.9 |
| C3. DEC + ECR (full) | ✓ Adjacent | ✓ Adjacent | 83.5 | 84.9 | 85.5 | 66.6 | 69.8 | 68.5 |
[1]Han, Yue-Hua, et al. "Towards more general video-based deepfake detection through facial component guided adaptation for foundation model." 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2025.
[2]Yan, Zhiyuan, et al. "Generalizing deepfake video detection with plug-and-play: Video-level blending and spatiotemporal adapter tuning." 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2025.
[3]Yermakov, Andrii, et al. "Deepfake detection that generalizes across benchmarks." 2026 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). IEEE, 2026.
[4]Arnab, Anurag, et al. "Vivit: A video vision transformer." 2021 IEEE/CVF international conference on computer vision (ICCV). IEEE, 2021.
[5]Nguyen, Dat, et al. "Vulnerability-aware spatio-temporal learning for generalizable deepfake video detection." 2025 IEEE/CVF International Conference on Computer Vision (ICCV). IEEE, 2025.
Response to Reviewer APAa's Weaknesses 1 and Question 1 & 2 (Continued)
W1.3 & Q2 A comparison against randomly paired perturbations.
To test whether TOP benefits from matching the same facial region across adjacent frames, we kept TAP and the first-frame TOP location fixed and shifted the second-frame patch center from to , where is a random spatial offset drawn from a predefined neighborhood. The original pairing outperformed this control in all six settings across DFD and DFDC (Table 3 here). This result shows that where TOP is applied in the second frame matters, beyond simply adding another perturbation type.
Table 3: Effect of Temporal Pairing and Feature Correspondence (AUC%).
| Setting | ECR perturbations | Frame pairing and feature correspondence | DFD | DFDC | ||||
|---|---|---|---|---|---|---|---|---|
| DeiT-B | ViT-B | Swin-B | DeiT-B | ViT-B | Swin-B | |||
| DEC only | None | No ECR calibration | 73.6 | 71.9 | 75.6 | 60.9 | 62.7 | 64.1 |
| TAP only | TAP | Matched locations | 77.6 | 79.5 | 78.3 | 63.5 | 66.6 | 65.8 |
| C2. Nonadjacent pairing | TAP + TOP | Nonadjacent frames | 75.5 | 70.0 | 77.7 | 60.0 | 65.3 | 61.9 |
| C1. Full ViGuardQ | TAP + TOP | Original | 83.5 | 84.9 | 85.5 | 66.6 | 69.8 | 68.5 |
| Random-offset TOP | TAP + TOP | Random | 74.5 | 73.3 | 76.0 | 61.5 | 63.2 | 65.0 |
Response to Reviewer APAa's Weaknesses 1 and Question 1 & 2
Thank you for the detailed comments. Hope we address all your concerns and please drop us follow-up questions if there are any.
W1 & Q1:
W1.1 Difference between ViGuardQ and Tao et al. (IJCAI 2025).
ViGuardQ differs from Tao et al. (IJCAI 2025) in three main aspects:
1) Target Model difference. Tao et al. primarily focus on quantizing traditional CNN-based deepfake detectors, whereas our evaluation targets widely adopted more advanced Transformer-based architectures, including DeiT, ViT, and Swin. Given this, /experimental performance comparison is inapplicable/ .
2) Quantization Protocol difference. Tao et al. adopt quantization-aware /training/ (QAT), which jointly optimizes the detector and quantization components. In contrast, ViGuardQ employs post-training quantization (PTQ), preserving the pretrained detector weights and requiring only a small calibration dataset. /We substantially reduces training costs and enable deployment on resource-constrained edge devices without retraining. /
3) Method difference. (1)Tao et al. focus on preserving spatial forgery textures in individual images, whereas ViGuardQ additionally considers temporal inconsistencies across video frames. (2) Specifically, our DEC combines within-frame decision gradients with cross-frame gradient differences to preserve spatiotemporal forgery evidence, while ECR aligns FP and quantized decision responses under temporally aligned and opposed perturbations.
Q1.1 Clarification on Matched Feature Location.
We first perform face detection, and spatial alignment on adjacent video frames. Corresponding facial regions are then mapped to the feature grids at each selected layer, providing the matched locations used to compute cross-frame differential gradients.