Welcome to the Cultural Captions Project!
Thank you for helping us! Computers are great at identifying objects in a photo (like a cup, a dog, or a building), but they are terrible at understanding human culture. We are building a dataset to teach AI the "why" behind an image. To do this, we need your help by sharing photos from your own life and explaining what they mean to someone who has never visited your country or experienced your culture.
Your task has three simple steps:
- Choose the Right Image
- Tag the Concepts
- Caption the Image
By uploading your data, you agree to it being licensed under CC-BY license. If you have further questions, reach out to us at wonders-team@googlegroups.com.
Please select your profile
Prefer not to log in? You can continue anonymously, but you won't be able to review, edit, or delete your submissions later. Also, at the end of the data collection campaign, we will use account information to create a list of all contributors and provide rewards to the best ones. For details, see our website https://team-wonders.github.io/.
Please select your country
The Cultural Captions Project: Teaching AI the about the culture behind the image
Step 1: Choose the Right Photo
Choose photos that tell a story about your community, habits, or traditions.
The "So What?" Test: Before you select a photo, ask yourself, "Would someone from a different country find this interesting or different from their own life?"
- Avoid: Generic landscapes, pretty sunsets, plain selfies, or standard macro shots of flowers. These look nice, but lack cultural depth.
- Choose: Local street food, traditional clothing, religious ceremonies, a typical daily commute, unique architecture, or social gatherings (like friends at a local pub, cafe, or market).
Step 2: Select the Main Concept
Before you write, you need to anchor your photo. What is the primary cultural idea you want to write about?
Select one concept from the following categories that best fits your image. If your concept is not present, you can add it by typing in the concept field.
- 🏛️ Buildings: (e.g., home decorations, local markets, evening hangouts, places of worship)
- 🎉 Celebrations, Sports and Events: (e.g., weddings, local festivals, religious events, national holidays)
- 👕 Clothing: (e.g., traditional dresses, contemporary clothing)
- 🍲 Food and Beverages: (e.g., a traditional dish, how a meal is served, dining etiquette)
- 🏘️ Instruments, Tools and Utensils: (e.g., musical instruments, tools, everyday-life objects)
Step 3: Write the Cultural Caption
Your goal: Answer this question: "If I showed this photo to a friend from another country, what would they need to know to understand what it means in your culture?"
Connect the Visuals to the Meaning: Cultural meaning is tied to physical details. Do not explain a cultural concept without first pointing out the exact things in the photo that represent it. If the texture of a bowl, a specific hand gesture, or the layout of a room is what makes this culturally significant, you must describe those physical details so the reader can see the connection.
Structure your caption like this:
- THE VISUAL EVIDENCE: What do we actually see in this photo that matters? Describe the specific physical details, objects, or actions that hold the cultural meaning. (e.g., "The photo shows people sitting cross-legged on woven mats around a low wooden table, sharing several small clay dishes of fermented cabbage...")
- THE CULTURAL NAME & CONTEXT: What is this called, and what is the context? Give the scene or items their specific cultural name. Who does this? When and where? Is it an everyday habit or a special occasion? (e.g., "...This is a traditional banchan spread, served alongside rice at almost every everyday meal in Korea.")
- THE "WHY": Why is this meaningful? What does this practice or scene represent about your community?
Step 1: Choose the Right Photo
Choose photos that tell a story about your community, habits, or traditions.
The "So What?" Test: Before you select a photo, ask yourself, "Would someone from a different country find this interesting or different from their own life?"
- Avoid: Generic landscapes, pretty sunsets, plain selfies, or standard macro shots of flowers. These look nice, but lack cultural depth.
- Choose: Local street food, traditional clothing, religious ceremonies, a typical daily commute, unique architecture, or social gatherings (like friends at a local pub, cafe, or market).
Technical requirements:
- The main concept should be clearly visible in the image in typical contexts where the chosen object would appear in real life. Other objects should preferably be visible too.
- Photos should be representative of your culture and population (e.g., people should look like the people from your ethnicity group).
- Images should NOT have empty backgrounds NOR watermarks.
- The complexity of the image will contribute to the interestingness of the description.
- You can blur all/selected faces before submitting your image. Once submitted, faces cannot be deblurred (you can still, however, re-upload the image if you want to).
If you don't have personal pictures for a concept, please use public images by providing their URL (instead of downloading them):
- Google Images
- When searching, click on "Tools" → "Usage rights" → "Creative Commons licenses"
- Click on an image to find the license in the "License details" panel on the right. Valid licenses are: Public domain, CC0, CC BY, CC BY-SA
- Unsplash
- After searching for a concept, click on "License" → "Free"
- Pexels
- All content is free
- Freepik
- Choose a "Free" license in the left panel after searching
Step 2: Select the Main Concept
Before you write, you need to anchor your photo. What is the primary cultural idea you want to write about?
Select one concept from the following categories that best fits your image. If your concept is not present, you can add it by typing in the concept field.
- 🏛️ Buildings: (e.g., home decorations, local markets, evening hangouts, places of worship)
- 🎉 Celebrations, Sports and Events: (e.g., weddings, local festivals, religious events, national holidays)
- 👕 Clothing: (e.g., traditional dresses, contemporary clothing)
- 🍲 Food and Beverages: (e.g., a traditional dish, how a meal is served, dining etiquette)
- 🏘️ Instruments, Tools and Utensils: (e.g., musical instruments, tools, everyday-life objects)
Additional concepts (optional)
Tag any other concept that is visible in the image.
Step 3: Write the Cultural Caption
Your goal: Answer this question: "If I showed this photo to a friend from another country, what would they need to know to understand what it means in your culture?"
Connect the Visuals to the Meaning: Cultural meaning is tied to physical details. Do not explain a cultural concept without first pointing out the exact things in the photo that represent it. If the texture of a bowl, a specific hand gesture, or the layout of a room is what makes this culturally significant, you must describe those physical details so the reader can see the connection.
Structure your caption like this:
- THE VISUAL EVIDENCE: What do we actually see in this photo that matters? Describe the specific physical details, objects, or actions that hold the cultural meaning. (e.g., "The photo shows people sitting cross-legged on woven mats around a low wooden table, sharing several small clay dishes of fermented cabbage...")
- THE CULTURAL NAME & CONTEXT: What is this called, and what is the context? Give the scene or items their specific cultural name. Who does this? When and where? Is it an everyday habit or a special occasion? (e.g., "...This is a traditional banchan spread, served alongside rice at almost every everyday meal in Korea.")
- THE "WHY": Why is this meaningful? What does this practice or scene represent about your community?
A description of the image with extensive, detailed visual and cultural information.
Descriptions must be objective: focus on how you would describe the image to someone who can't see it, without your own opinions/speculations.
Writing guidelines:
- Avoid redundant phrases such as "This is an image of." It is okay to specify the medium, such as "A painting of."
- Check spelling, grammar, and punctuation.
- Relevant text visible in the image should be completely transcribed within "quotation marks." For example: A woman holding a "Stop war" sign.
- Start the description with the elements that are critical to understanding the image — typically the first thing that strikes you.
- Do not mention people's names. For example, "Taylor Swift" → "A blonde, female singer."
Describing people and things:
- When particular features are immediately noticeable, describe them (e.g., "a person in a wheelchair").
- Describe age using terminology such as "baby," "child," "young person," "adult," "older person," etc.
- When relevant, describe skin tone using non-ethnic terms such as "light-skinned" or "dark-skinned" when clearly visible.
- For animals, use the species or breed name (e.g., "capybara" or "husky") rather than generic terms (e.g., "dog").
- You can mention names of places, but also include their common name — especially when not widely known. For example: "in front of the London Eye, a renowned ferris wheel in London."
AI Description
A state-of-the-art AI model has also been asked to generate a similar description for this image.
Your feedback (Yes/No) helps us understand how well current AI models capture your cultural context.
Model is randomly assigned when the caption is generated.
When enabled, you can submit with only image and main concept.
👤 Your Profile
📊 Comparative Statistics
How you compare against the global userbase and the communities you contribute to.
Loading statistics…
📈 Your Contribution Metrics
Loading metrics…
Loading…
📋 Your Contributed Data
Loading your data, please wait ...