Since our CloudAPI and Self Hosted API receive single images for recognition through the requests you send, it is necessary to make sure the images you're sending are of sufficient quality.
During the scanning process, one of the first steps is the detection of the document on the image which requires the document edges to be visible. To aid document detection, we recommend leaving a small margin around the document and preferably having a contrasting background. More details on required image quality and margin size can be found in the documentation.
In some cases, leaving a margin around the document might be impossible (e.g. the images you're sending for recognition are already cropped tightly around the document). In those cases, we provide a request parameter which will let the API skip document detection and go straight to the recognition process. To enable this behavior, it is necessary to set the following request parameter:
If image acquisition still causes problems in your solution, we recommend looking into our Image Capture In-browser SDK, a WebAssembly based solution used on your frontend which captures the best possible frames from a video stream from the user's camera and sends them off to your backend (either SH API or CloudAPI).