The People Counting model determines the number of different people visible within a given image or video frame.
The model does not rely on face detection. Individuals will be counted by the model regardless of face visibility, ensuring accurate detection even when only partial body parts are shown.
If a person appears multiple times in the frame, such as reflections in mirrors or image collages, the model will correctly count them only once.
The model has been trained to distinguish real people from non-human representations, explicitly excluding mannequins, dolls, statues and illustrated figures like cartoons or drawings.
The response is structured into 6 distinct classes, based on the number of unique people visible:
| Class name | Description |
|---|---|
| 0 | No people visible in the frame |
| 1 | One person in the frame |
| 2 | Two people in the frame |
| 3 | Three people in the frame |
| 4 | Four people in the frame |
| 5+ | Five or more people in the frame |
For video inputs, the model will analyze each frame individually and return the number of people visible within each frame.
If you haven't already, create an account to get your own API keys.
This model is currently gated. Please get in touch for access and implementation details.
This model is currently gated. Please get in touch for access and implementation details.
See our full list of Image/Video models for details on other filters and checks you can run on your images and videos. You might also want to check our Text models to moderate text-based content: messages, reviews, comments, usernames...
Was this page helpful?