# PIXELMOD: Improving Soft Moderation of Visual Misleading Information on Twitter

arXiv: https://arxiv.org/abs/2407.20987 · HTML: https://mlsanity.com/paper/2407.20987/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 86 references that are in the map.
- First posted on arXiv 30 Jul 2024.

## Builds on (most cited first, 28 of 28)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Rethinking the Inception Architecture for Computer Vision](https://mlsanity.com/paper/1512.00567/) (2015)
- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [VisualBERT: A Simple and Performant Baseline for Vision and Language](https://mlsanity.com/paper/1908.03557/) (2019)
- [Learning Rich Features for Image Manipulation Detection](https://mlsanity.com/paper/1805.04953/) (2018)
- [Open-Domain, Content-based, Multi-modal Fact-checking of Out-of-Context Images via Online Resources](https://mlsanity.com/paper/2112.00061/) (2021)
- [On the Origins of Memes by Means of Fringe Web Communities](https://mlsanity.com/paper/1805.12512/) (2018)
- [BadHash: Invisible Backdoor Attacks against Deep Hashing with Clean Label](https://mlsanity.com/paper/2207.00278/) (2022)
- [COSMOS: Catching Out-of-Context Misinformation with Self-Supervised Learning](https://mlsanity.com/paper/2101.06278/) (2021)
- [On the Evolution of (Hateful) Memes by Means of Multimodal Contrastive Learning](https://mlsanity.com/paper/2212.06573/) (2022)
- [Understanding and Detecting Hateful Content using Contrastive Learning](https://mlsanity.com/paper/2201.08387/) (2022)
- [Images and Misinformation in Political Groups: Evidence from WhatsApp in India](https://mlsanity.com/paper/2005.09784/) (2020)
- ["I Won the Election!": An Empirical Analysis of Soft Moderation Interventions on Twitter](https://mlsanity.com/paper/2101.07183/) (2021)
- [TROLLMAGNIFIER: Detecting State-Sponsored Troll Accounts on Reddit](https://mlsanity.com/paper/2112.00443/) (2021)
- [Understanding the Use of Fauxtography on Social Media](https://mlsanity.com/paper/2009.11792/) (2020)
- [LAMBRETTA: Learning to Rank for Twitter Soft Moderation](https://mlsanity.com/paper/2212.05926/) (2022)
- [Hiding in Plain Sight: A Longitudinal Study of Combosquatting Abuse](https://mlsanity.com/paper/1708.08519/) (2017)
- [VoterFraud2020: a Multi-modal Dataset of Election Fraud Claims on Twitter](https://mlsanity.com/paper/2101.08210/) (2021)
- [Dial One for Scam: A Large-Scale Analysis of Technical Support Scams](https://mlsanity.com/paper/1607.06891/) (2016)
- [Toxicity in the Decentralized Web and the Potential for Model Sharing](https://mlsanity.com/paper/2204.12709/) (2022)
- [Identifying Misinformation from Website Screenshots](https://mlsanity.com/paper/2102.07849/) (2021)
- [Tiplines to Combat Misinformation on Encrypted Platforms: A Case Study of the 2019 Indian Election on WhatsApp](https://mlsanity.com/paper/2106.04726/) (2021)
- [Coordinated through aWeb of Images: Analysis of Image-based Influence Operations from China, Iran, Russia, and Venezuela](https://mlsanity.com/paper/2206.03576/) (2022)
- [A First Look at COVID-19 Messages on WhatsApp in Pakistan](https://mlsanity.com/paper/2011.09145/) (2020)

## Abstract (arXiv)

Images are a powerful and immediate vehicle to carry misleading or outright false messages, yet identifying image-based misinformation at scale poses unique challenges. In this paper, we present PIXELMOD, a system that leverages perceptual hashes, vector databases, and optical character recognition (OCR) to efficiently identify images that are candidates to receive soft moderation labels on Twitter. We show that PIXELMOD outperforms existing image similarity approaches when applied to soft moderation, with negligible performance overhead. We then test PIXELMOD on a dataset of tweets surrounding the 2020 US Presidential Election, and find that it is able to identify visually misleading images that are candidates for soft moderation with 0.99% false detection and 2.06% false negatives.