Hide and Recognize Your Privacy Image
Bibliographic record
Abstract
Recent studies have demonstrated that deep neural networks show excellent performance in information hiding. Considering the tremendous progress that deep learning has made in image recognition, we explore whether neural networks can recognize invisible private images hidden in cover images. In this article, we propose a method for image recognition in the covert domain using neural networks. Our target is to hide an image inside another image with minimal visual quality loss, while at the same time, the hidden image can be recognized correctly without being recovered. In the proposed system, the hiding and recognition of secret images are all performed by neural networks. The hiding network and the recognition network are designed to specifically work as a pair. We design and jointly train preparation, hiding, and recognition networks, where given a cover and a secret image, the preparation network reduces redundant information of the secret image, the hiding network produces a stego image that is visually indistinguishable from the cover image, and the PSNR and SSIM reach 38.5 dB and 0.991 on the MNIST & CIFAR-10 dataset and 41.8 dB and 0.995 on the CelebA & Scene dataset, respectively. The recognition network can correctly identify the secret image inside the stego image which reaches 98.3% recognition accuracy on MNIST dataset and 91.6% recognition accuracy on CelebA dataset in the covert domain, less than 1% recognition decrease compared with direct recognition. In summary, our approach can successfully identify the secret image without revealing its content. Across various datasets, both the classification accuracy and the invisibility of private images are consistently satisfactory.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.004 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".