On the Effectiveness of Dataset Alignment for Fake Image Detection

• URL: http://arxiv.org/abs/2410.11835v1
• Date: Tue, 15 Oct 2024 17:58:07 GMT
• Title: On the Effectiveness of Dataset Alignment for Fake Image Detection
• Authors: Anirudh Sundara Rajan, Utkarsh Ojha, Jedidiah Schloesser, Yong Jae Lee,
• Abstract summary: A good detector should focus on the generative models fingerprints while ignoring image properties such as semantic content, resolution, file format, etc. In this work, we argue that in addition to these algorithmic choices, we also require a well aligned dataset of real/fake images to train a robust detector. For the family of LDMs, we propose a very simple way to achieve this: we reconstruct all the real images using the LDMs autoencoder, without any denoising operation. We then train a model to separate these real images from their reconstructions.
• Score: 28.68129042301801
• License:
• Abstract: As latent diffusion models (LDMs) democratize image generation capabilities, there is a growing need to detect fake images. A good detector should focus on the generative models fingerprints while ignoring image properties such as semantic content, resolution, file format, etc. Fake image detectors are usually built in a data driven way, where a model is trained to separate real from fake images. Existing works primarily investigate network architecture choices and training recipes. In this work, we argue that in addition to these algorithmic choices, we also require a well aligned dataset of real/fake images to train a robust detector. For the family of LDMs, we propose a very simple way to achieve this: we reconstruct all the real images using the LDMs autoencoder, without any denoising operation. We then train a model to separate these real images from their reconstructions. The fakes created this way are extremely similar to the real ones in almost every aspect (e.g., size, aspect ratio, semantic content), which forces the model to look for the LDM decoders artifacts. We empirically show that this way of creating aligned real/fake datasets, which also sidesteps the computationally expensive denoising process, helps in building a detector that focuses less on spurious correlations, something that a very popular existing method is susceptible to. Finally, to demonstrate just how effective the alignment in a dataset can be, we build a detector using images that are not natural objects, and present promising results. Overall, our work identifies the subtle but significant issues that arise when training a fake image detector and proposes a simple and inexpensive solution to address these problems.

Related papers

• Zero-Shot Detection of AI-Generated Images [54.01282123570917]
  We propose a zero-shot entropy-based detector (ZED) to detect AI-generated images. Inspired by recent works on machine-generated text detection, our idea is to measure how surprising the image under analysis is compared to a model of real images. ZED achieves an average improvement of more than 3% over the SoTA in terms of accuracy.
  arXiv  Detail & Related papers  (2024-09-24T08:46:13Z)
• Contrasting Deepfakes Diffusion via Contrastive Learning and Global-Local Similarities [88.398085358514]
  Contrastive Deepfake Embeddings (CoDE) is a novel embedding space specifically designed for deepfake detection. CoDE is trained via contrastive learning by additionally enforcing global-local similarities.
  arXiv  Detail & Related papers  (2024-07-29T18:00:10Z)
• Detecting Generated Images by Real Images Only [64.12501227493765]
  Existing generated image detection methods detect visual artifacts in generated images or learn discriminative features from both real and generated images by massive training. This paper approaches the generated image detection problem from a new perspective: Start from real images. By finding the commonality of real images and mapping them to a dense subspace in feature space, the goal is that generated images, regardless of their generative model, are then projected outside the subspace.
  arXiv  Detail & Related papers  (2023-11-02T03:09:37Z)
• AntifakePrompt: Prompt-Tuned Vision-Language Models are Fake Image Detectors [24.78672820633581]
  Deep generative models can create remarkably fake images while raising concerns about misinformation and copyright infringement. Deepfake detection technique is developed to distinguish between real and fake images. We propose a novel approach called AntifakePrompt, using Vision-Language Models and prompt tuning techniques.
  arXiv  Detail & Related papers  (2023-10-26T14:23:45Z)
• SISL:Self-Supervised Image Signature Learning for Splicing Detection and Localization [11.437760125881049]
  We propose self-supervised approach for training splicing detection/localization models from frequency transforms of images. Our proposed model can yield similar or better performances on standard datasets without relying on labels or metadata.
  arXiv  Detail & Related papers  (2022-03-15T12:26:29Z)
• Enhancing Low-Light Images in Real World via Cross-Image Disentanglement [58.754943762945864]
  We propose a new low-light image enhancement dataset consisting of misaligned training images with real-world corruptions. Our model achieves state-of-the-art performances on both the newly proposed dataset and other popular low-light datasets.
  arXiv  Detail & Related papers  (2022-01-10T03:12:52Z)
• DA-FDFtNet: Dual Attention Fake Detection Fine-tuning Network to Detect Various AI-Generated Fake Images [21.030153777110026]
  It has been much easier to create fake images such as "Deepfakes" Recent research has introduced few-shot learning, which uses a small amount of training data to produce fake images and videos more effectively. In this work, we propose Dual Attention Fine-tuning Network (DA-tNet) to detect the manipulated fake face images.
  arXiv  Detail & Related papers  (2021-12-22T16:25:24Z)
• Data Augmentation for Object Detection via Differentiable Neural Rendering [71.00447761415388]
  It is challenging to train a robust object detector when annotated data is scarce. Existing approaches to tackle this problem include semi-supervised learning that interpolates labeled data from unlabeled data. We introduce an offline data augmentation method for object detection, which semantically interpolates the training data with novel views.
  arXiv  Detail & Related papers  (2021-03-04T06:31:06Z)
• Identifying Invariant Texture Violation for Robust Deepfake Detection [17.306386179823576]
  We propose the Invariant Texture Learning framework, which only accesses the published dataset with low visual quality. Our method is based on the prior that the microscopic facial texture of the source face is inevitably violated by the texture transferred from the target person.
  arXiv  Detail & Related papers  (2020-12-19T03:02:15Z)
• What makes fake images detectable? Understanding properties that generalize [55.4211069143719]
  Deep networks can still pick up on subtle artifacts in doctored images. We seek to understand what properties of fake images make them detectable. We show a technique to exaggerate these detectable properties.
  arXiv  Detail & Related papers  (2020-08-24T17:50:28Z)
• FDFtNet: Facing Off Fake Images using Fake Detection Fine-tuning Network [19.246576904646172]
  We propose a light-weight fine-tuning neural network-based architecture called FaketNet. Our approach aims to reuse popular pre-trained models with only a few images for fine-tuning to effectively detect fake images. Our tNet achieves an overall accuracy of 9029% in detecting fake images generated from the GANs-based dataset.
  arXiv  Detail & Related papers  (2020-01-05T16:04:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.