- Rating
- 3(+1)
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
This paper proposes a detector-guided feature synthesis framework to improve the cross-generator generalization of AI-generated image detection. To reduce shortcut learning caused by pixel-level synthesis artifacts, the method constructs hard fake features by interpolating aligned real and fake representations. A lightweight mixing coefficient generator is then optimized adversarially using detector feedback to produce more challenging, pair-adaptive fake features. Additionally, a diversity constraint is introduced to prevent coefficient collapse during training.
Critical strengths
-
This paper is well organized and easy to follow.
-
Sufficient discussions in the paper to help researchers understand the method: detailed comparison experiments and method discussions.
Critical weaknesses
- I have doubts about the core assumption of this paper that "the mixture of real and generated features still belongs to the generated category." Simple linear feature interpolation cannot guarantee that the mixed features correspond to a real generated image, let alone guarantee that it still retains reliable generation traces. Therefore, directly labeling all mixed features as Fake lacks sufficient theoretical basis and may introduce label ambiguity.
- The fairness of the comparative experiments is questionable. The authors directly used the pre-trained weights officially released by each comparative method, but these models were not obtained under the same dataset and training protocol, so the performance differences may not entirely come from the methods themselves.
- The authors emphasize that pixel space mixing may introduce additional artifacts, but this paper still relies on pixel space-aligned images constructed by DDA for training. This means that the method has not eliminated the bias that pixel-space data construction may bring, and there is still a certain gap between its core motivation and actual implementation.
- The current ablation experiments lack validation on different backbones. It is recommended that the authors further analyze the performance of this method under different feature extraction networks to prove that its effectiveness does not depend on specific pre-trained representations.
Questions for authors
Please refer to the Weaknesses section above for detailed questions and concerns.
- What justifies assigning the Fake label to interpolated real–fake features? How can the authors ensure that these features retain meaningful forensic traces rather than introducing label ambiguity?
- Since the compared methods use official checkpoints trained on different datasets and under different protocols, how can the authors ensure that the performance comparisons are fair?
- The paper argues that pixel-space synthesis introduces spurious artifacts, yet the training data still rely on DDA-generated aligned images. How does the proposed method avoid inheriting these artifacts and the associated biases?
- The experiments only use DINOv2 as the backbone. Can the authors clarify whether the proposed feature synthesis strategy is expected to generalize to other backbone architectures?
4 comments
W3 & Q3. Pixel-mixing shortcuts and inherited biases
-
Pixel-blending artifacts can become shortcuts when blended images are labeled fake. To avoid introducing these cues during hard-negative synthesis, we encode each real image and its DDA-aligned reconstruction separately, then pass their interpolated features directly to the classifier.
-
To assess sensitivity to pixel blending, we compared fake detection rates on paired Recon and PixelMix inputs. Each detector was evaluated on 1,000 pairs of VAE reconstructions and their pixel-space blends with the source images. The gap is the fake detection rate on PixelMix minus that on Recon, in percentage points (pp).
Table R2. Fake detection rates on Recon and PixelMix
Detector gap (pp) PixelMix-trained +27.3 FeatureMix-trained +4.7 The smaller gap for the FeatureMix-trained detector suggests that feature mixing reduces the detector’s reliance on pixel-blending cues as a shortcut for identifying fakes.
W4 & Q4. Transfer to a CLIP backbone
-
To test whether our framework transfers beyond DINOv2, we trained a CLIP variant on COCO-SD2 and evaluated it on the same nine benchmarks. UnivFD, C2P-CLIP, and DRCT use the matched-data results from Table R1. FatFormer uses its official checkpoint.
Table R3. Accuracy of CLIP-based detectors across nine benchmarks
Method AIGCDB DRCT-2M WildRF ForenSynths EvalGEN Synthbuster SynthWildX Chameleon GenImage Mean ACC (%) UnivFD 78.53 86.85 56.00 67.41 88.87 62.30 56.25 61.25 85.12 71.40 FatFormer 74.78 51.95 55.60 84.52 48.93 53.62 53.38 58.03 61.03 60.20 C2P-CLIP 66.15 69.85 59.32 50.95 53.80 49.36 54.25 56.73 74.68 59.46 DRCT 78.74 85.34 58.87 68.83 73.93 70.65 60.71 63.51 81.56 71.35 Ours (CLIP) 86.61 96.76 80.71 71.55 86.97 65.83 80.07 64.51 90.84 80.43 Our CLIP variant achieves 80.43% mean ACC, 9.03 percentage points above the strongest baseline in this comparison. These results support the applicability of our feature synthesis framework to CLIP.
Thank you for your detailed reply, which resolved my question, so I am increasing my score.
Thank you for reading our response and updating your assessment. We are glad our reply addressed your question, and we appreciate your constructive feedback.
Dear Reviewer #pb62, Thank you for taking the time to review our paper and for your constructive feedback. We hope the clarifications and additional results below address your concerns.
W1 & Q1. Justification for fake labels on mixed features
To explain which cues can survive interpolation, we consider the feature difference between aligned real and reconstructed images. For each pair, zm−zr=(1−λ)(zf−zr). If a cue is expressed as a linear contrast between the endpoints, the mixed feature preserves the sign of that contrast relative to the real endpoint and scales its magnitude by 1−λ. The mixed feature can therefore expose the detector to a weaker version of the same cue.
We label the mixed features as fake to train the detector on weaker forensic cues. To limit label ambiguity, we set λmax=0.8, which keeps each mixed feature at least 20% of the original pairwise distance from its real endpoint. We also supervise the original real image and its reconstruction with their respective labels.
In Table 3, the fixed-coefficient feature synthesis variant (λ0=0.65) achieves 86.10% mean ACC across nine benchmarks, compared with 82.64% for the original no-synthesis baseline.
W2 & Q2. Baseline comparison on matched training data
To control for training-data differences, we retrained NPR, UnivFD, SAFE, C2P-CLIP, AIDE, and DRCT on the same COCO-SD2 dataset as our method. It consists of real MS COCO images paired with their Stable Diffusion 2.0 VAE reconstructions. The AlignedForensics and DDA results in the original submission already come from models we trained on this dataset. FatFormer uses the official checkpoint because its training code was unavailable to us.
We evaluated all detectors on the same nine benchmarks. Mean is the unweighted average of the nine benchmark-level accuracies. AIGCDB denotes AIGCDetectionBenchmark.
Table R1. Accuracy on nine benchmarks with matched training data
Our method achieves the highest reported accuracy on all nine benchmarks. Its mean ACC is 89.14%, which is 9.45 percentage points higher than DDA (79.69%), the strongest baseline trained on the same data. The accuracy advantage therefore persists when the training dataset is held fixed.