Skripsi
PROTEKSI CITRA FOTO KPM MAHASISWA FASILKOM UNSRI DARI DEEPFAKE DENGAN CMUA-WATERMARK
The development of technology and information provides great benefits for information exchange, but also opens up opportunities for misuse of information and the spread of harmful hoaxes. The deepfake phenomenon, which is the manipulation of images and videos using Artificial Intelligence (AI), poses a serious threat to security and privacy. This research focuses on deepfakes of human face images, especially those obtained from the Sriwijaya University (UNSRI) website, which can be exploited without verification. Commonly used passive defense approaches to detect deepfakes have limitations because the modified images remain at risk of being widely spread on the internet. Therefore, this research proposes an active defense approach using Adversarial Watermarking techniques. The proposed method, Cross-Model Universal Adversarial Watermark (CMUA-Watermark), protects thousands of face images from various deepfake models simultaneously. The results show that CMUA-Watermark provides a high protection success rate with satisfactory performance on various datasets. However, the HiSD deepfake model shows a decrease in performance against the KPM photo image dataset of Fasilkom UNSRI students. Therefore, the use of CMUA-Watermark on older versions of UNSRI websites is recommended to prevent deepfake exploitation and maintain the security and privacy of student facial images.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2307003787 | T127121 | T1271212023 | Central Library (REFERENS) | Available but not for loan - Not for Loan |
No other version available