in transitu, 2021-Now

in transitu is an ongoing performative photo project on Instagram about how algorithms understand gender.

Each week, I document my gender transition by uploading an image of my bare chest on Instagram. Since the platform doesn't allow female nipples, the question is: when does my body fit the social media site's definition of womanhood?

After running the project for 10 weeks, I decided to give my followers an additional insight into how gender classification algorithms work. While we await Instagram censorship, I provide results from various publicly available gender classification algorithms. This gives a continuous look into how computers attempt to classify human gender.

The project acts as a sort of reverse engineering of Instagram's moderation protocols. In using my own transgender body, I show how inconsistent the inner workings of gender classification algorithms are, whether used for censorship or offered as a software service.

On the 30th of March, in transitu had its first image removed from Instagram. It didn't last long, however, because the image was put back up a week later. I hadn't made a formal complaint over Instagram's decisions. They simply decided that the image did not, in fact, contain female nipples.

As of 17th of September 2025, the project is in a state of censorship limbo. Some images have been completely removed, while others are shadowbanned, have their reach reduced or are marked as sensitive content. Yet the Instagram profile continues to exist as a whole.

Through these discoveries, in transitu continues to bring new perspectives to discussions surrounding gender, "artificial intelligence", algorithms, internet culture, and censorship.

If you want a thorough recounting of all the things that have happened in the course of the project, you can read my freely available article over on Patreon.

When I reached post number 200, I changed the AI classification software used for the project. You can read more about that here.

All the images are taken in my own bedroom with my smartphone. The gender and NSFW classification results come from various sources.

Microsoft results are from LLaVa:7b via Ollama. Meta results are from Llama3.2-vision:11b via Ollama. Google results are from Gemma3:4b via Ollama. Alibaba results are from Qwen3-vl:8b-instruct via Ollama. Amazon results are from the Rekognition API.