Xiaochuan Jin

Zheling Meng, Bo Peng, Xiaochuan Jin et al.

Text-to-image diffusion models have been demonstrated with undesired generation due to unfiltered large-scale training data, such as sexual images and copyrights, necessitating the erasure of undesired concepts. Most existing methods focus on modifying the generation probabilities conditioned on the texts containing target concepts. However, they fail to guarantee the desired generation of texts unseen in the training phase, especially for the adversarial texts from malicious attacks. In this paper, we analyze the erasure task and point out that existing methods cannot guarantee the minimization of the total probabilities of undesired generation. To tackle this problem, we propose Dark Miner. It entails a recurring three-stage process that comprises mining, verifying, and circumventing. This method greedily mines embeddings with maximum generation probabilities of target concepts and more effectively reduces their generation. In the experiments, we evaluate its performance on the inappropriateness, object, and style concepts. Compared with the previous methods, our method achieves better erasure and defense results, especially under multiple adversarial attacks, while preserving the native generation capability of the models. Our code will be available on GitHub.

Zheling Meng, Bo Peng, Xiaochuan Jin et al.

Text-to-image diffusion models have demonstrated the underlying risk of generating various unwanted content, such as sexual elements. To address this issue, the task of concept erasure has been introduced, aiming to erase any undesired concepts that the models can generate. Previous methods, whether training-based or training-free, have primarily focused on the input side, i.e., texts. However, they often suffer from incomplete erasure due to limitations in the generalization from limited prompts to diverse image content. In this paper, motivated by the notion that concept erasure on the output side, i.e., generated images, may be more direct and effective, we propose Concept Corrector. It checks target concepts based on visual features provided by final generated images predicted at certain time steps. Further, it incorporates Concept Removal Attention to erase generated concept features. It overcomes the limitations of existing methods, which are either unable to remove the concept features that have been generated in images or rely on the assumption that the related concept words are contained in input prompts. In the whole pipeline, our method changes no model parameters and only requires a given target concept as well as the corresponding replacement content, which is easy to implement. To the best of our knowledge, this is the first erasure method based on intermediate-generated images, achieving the ability to erase concepts on the fly. The experiments on various concepts demonstrate its impressive erasure performance.

Bo Peng, Zichuan Wang, Sheng Yu et al.

Deep learning based face-swap videos, widely known as deepfakes, have drawn wide attention due to their threat to information credibility. Recent works mainly focus on the problem of deepfake detection that aims to reliably tell deepfakes apart from real ones, in an objective way. On the other hand, the subjective perception of deepfakes, especially its computational modeling and imitation, is also a significant problem but lacks adequate study. In this paper, we focus on the visual realism assessment of deepfakes, which is defined as the automatic assessment of deepfake visual realism that approximates human perception of deepfakes. It is important for evaluating the quality and deceptiveness of deepfakes which can be used for predicting the influence of deepfakes on Internet, and it also has potentials in improving the deepfake generation process by serving as a critic. This paper prompts this new direction by presenting a comprehensive benchmark called DREAM, which stands for Deepfake REalism AssessMent. It is comprised of a deepfake video dataset of diverse quality, a large scale annotation that includes 140,000 realism scores and textual descriptions obtained from 3,500 human annotators, and a comprehensive evaluation and analysis of 16 representative realism assessment methods, including recent large vision language model based methods and a newly proposed description-aligned CLIP method. The benchmark and insights included in this study can lay the foundation for future research in this direction and other related areas.