Face Mask Detection API - Maskerizer (also known as face mask recognition API) is a cross browsers REST API which get a JSON input with a still photo (as base64 encoded string) or an url of the image and returns a JSON string which contains predictions with certain amount of probability (filtered for output with minimum 50%), bounding boxes of detected face mask(s) put on with its top, left, width, height values. For one still photo the API may return multiple predictions with different probability scores of detected face masks put on. Our pricing packages count the predictions, so for one request, multiple predictions may be counted. We filter the results of predictions so we display only the predictions with a probability score higher than 50%. Of course, there are some limitations in order to get a higher accuracy. We recommend properly exposed, unobstructed JPEG photos at 1920x1080 (full HD resolution) where ratio between height of bounding box and height of entire picture should be at least 1:15. For ratios like 1:16, 1:17, 1:18 and so on the accuracy is lower and the AI algorithm may not see the face massk and the prediction score will be low. We do not store pictures. Also, the quality and the angles of the camera are very important and it contribute to a higher detecting accuracy. It should have varifocal lenses, high shutter speed, good infrared lighting beam, full HD resolution.
Allthough this Face Mask Detection API (currently we do not offer a face mask recognition sdk) is intended for software development and therefore developers, we have also here an face mask detection online application that may be used to check the input and output JSONs of the API. The necessary steps are written below, basically for this real time face mask detection or recognition API you send an authorized POST request in JSON format to the API endpoint and you get as JSON response the output as described on our website.