Frequently Asked Questions

This page lists some frequently asked questions of AnySee.

Q1. Where can I find the API specification?

Please find it here.

Q2. How can I get an API key?

Please refer to this page.

Q3. How can I contact support?

For customers with a paid order, please use the Jira support page to create a supporting ticket.

For customers without a paid order, please use the discussion page or send an email to the following mail address.

For customers who want to place a paid order, use the contact form.

Q4. Does the service support facial recognition with masks?

Yes, with the currently available model JCV_FACE_K25000.

Q5. Is there a limit on the number of face entities that could be registered?

No, there is no limit on the number of face entities.

Q6. Is there a limit on the number of face entities in the search API?

No, there is no limit on the number of face entities in the search API.
But we recommend limiting the search range to get better performance in speed and accuracy.

Q7. What will happen if the API quota is exceeded?

You can still use the API service as usual. But you might be charged more compared to a standard rate.

Q8. What are the recommended input devices?

AnySee does not rely on a specific type of device. Any images that fulfill the input image requirements are accepted.

Q9. Why do I get an extremely low similarity score for the same person or cannot get a valid search result?

This is usually caused by an image in the wrong orientation. Please use EXIF viewers to check. Refer to this section for more details.

Q10. What is the recommended threshold?

We highly recommend determining your customized threshold value. For reference only, 0.90-0.95 is a frequently used value range. Please check this section for more details.

Q11. Why do I receive the 415 Unsupported Media Type error?

For POST, PATCH, and PUT methods, the API gateway applies a validation checker on the Content-Type header. The only accepted value is application/json. Please make sure you do not use other values like application/json; charset=utf-8.

Q12. Does the service store images?

In AnySee, user-uploaded image files or image data will not be stored anywhere in any condition. Image data in base64 encoding is disposed after detection or feature extraction. For the investigation purpose, users should prepare their own image storage database.

If you want a service with a photo storage function, please refer to the Identity Manager service.

Q13. How good is the facial recognition model?

Here is a comparison with some popular open-source facial recognition models.

ROC Curve against open-source models

ROC Curve against open-source models

As is explained in the graph, AnySee's model (JCV_FACE_K25000) is way better than most open-source models.