Do you use image metadata to moderate images?
No we don't. We know image metadata is not reliable as it is easily changed or removed. Just as humans will not need metadata to determine what an image contains, so do we. You can name your image 'safe.jpg' or 'nsfw.jpg' or whatever, that won't change the result.
The same holds true for the EXIF data, or any header / type information we might get from an image URL. The only data that we use for moderation are the actual pixel values.
Other frequent questions
- What is a model?
- What are the ways to send an image to the API?
- What image types are supported?
- What are the acceptable / recommended image dimensions?
- How do you process GIF images with multiple frames (animated GIFs)?
- Is it possible to perform bulk image moderation / to submit images in batch? How many images can be submitted at once?
- Can you detect duplicate images, or image spam?
- Is it possible to customize the Image Moderation API to my needs?
- Can you improve the moderation results based on my reports of incorrectly classified or mislabeled images?