# ImagiNet: A Multi-Content Benchmark for Synthetic Image Detection

arXiv: https://arxiv.org/abs/2407.20020 · HTML: https://mlsanity.com/paper/2407.20020/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Jul 2024.

## Built on by (most cited first, 5 of 5)

- [Few-Shot Synthetic Image Attribution: Identifying Unseen Generators with Limited Samples](https://mlsanity.com/paper/2509.25682/) (2025)
- [UniAIDet: A Unified and Universal Benchmark for AI-Generated Image Content Detection and Localization](https://mlsanity.com/paper/2510.23023/) (2025)
- [Generalized Design Choices for Deepfake Detectors](https://mlsanity.com/paper/2511.21507/) (2025)
- [Redefining Generalization in Visual Domains: A Two-Axis Framework for Fake Image Detection with FusionDetect](https://mlsanity.com/paper/2510.05740/) (2025)
- [Trustworthy synthetic visual media: Evidence across the media lifecycle](https://mlsanity.com/paper/2609.34232/) (2026)

## Abstract (arXiv)

Recent generative models produce images with a level of authenticity that makes them nearly indistinguishable from real photos and artwork. Potential harmful use cases of these models, necessitate the creation of robust synthetic image detectors. However, current datasets in the field contain generated images with questionable quality or have examples from one predominant content type which leads to poor generalizability of the underlying detectors. We find that the curation of a balanced amount of high-resolution generated images across various content types is crucial for the generalizability of detectors, and introduce ImagiNet, a dataset of 200K examples, spanning four categories: photos, paintings, faces, and miscellaneous. Synthetic images in ImagiNet are produced with both open-source and proprietary generators, whereas real counterparts for each content type are collected from public datasets. The structure of ImagiNet allows for a two-track evaluation system: i) classification as real or synthetic and ii) identification of the generative model. To establish a strong baseline, we train a ResNet-50 model using a self-supervised contrastive objective (SelfCon) for each track which achieves evaluation AUC of up to 0.99 and balanced accuracy ranging from 86% to 95%, even under conditions that involve compression and resizing. The provided model is generalizable enough to achieve zero-shot state-of-the-art performance on previous synthetic detection benchmarks. We provide ablations to demonstrate the importance of content types and publish code and data.